CA1085514A - Data compression system - Google Patents
Data compression systemInfo
- Publication number
- CA1085514A CA1085514A CA287,295A CA287295A CA1085514A CA 1085514 A CA1085514 A CA 1085514A CA 287295 A CA287295 A CA 287295A CA 1085514 A CA1085514 A CA 1085514A
- Authority
- CA
- Canada
- Prior art keywords
- character
- digital
- bits
- boundary
- pattern
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Expired
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T9/00—Image coding
- G06T9/20—Contour coding, e.g. using detection of edges
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V30/00—Character recognition; Recognising digital ink; Document-oriented image-based pattern recognition
- G06V30/10—Character recognition
- G06V30/18—Extraction of features or characteristics of the image
- G06V30/182—Extraction of features or characteristics of the image by coding the contour of the pattern
- G06V30/1823—Extraction of features or characteristics of the image by coding the contour of the pattern using vector-coding
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N1/00—Scanning, transmission or reproduction of documents or the like, e.g. facsimile transmission; Details thereof
- H04N1/41—Bandwidth or redundancy reduction
- H04N1/411—Bandwidth or redundancy reduction for the transmission or storage or reproduction of two-tone pictures, e.g. black and white pictures
- H04N1/413—Systems or arrangements allowing the picture to be reproduced without loss or modification of picture-information
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V30/00—Character recognition; Recognising digital ink; Document-oriented image-based pattern recognition
- G06V30/10—Character recognition
Abstract
Abstract of the Disclosure A data compression system is disclosed in which the left-hand boundary of a character is developed in the form of a sequence of Freeman direction codes, the codes being stored in digital form within a processor. the number of binary data bits required to define the character using different criteria is then generated and compared to determine which criteria defines the character in the minimum amount of binary data bits.
Description
Back~round of the Invention Modern day data processing of document6, such as checks, have included proposed systems in which the digital image of the document i8 recorded on a recording medium such as tape for transmission to a remote station. Prior methods of recording such images consisted of scanning the document and storing on a recording medium such as magnetic tape the pattern of the image in the form of binary data bits for trans-mission to a remote station where the image of the document is reproduced utilizing the recorded data. This has resulted in an excessively large amount of binary data which must be stored for further processing. In order to reduce this data, character thinning proces~es have been used to reduce the prlnted characters or other shapes on the document to their ba8ic outline. Examples of such a process may be found in United State~ patents Nos. 3,196,398; 3,975,709; 3,846,754, and 3,863,218. While the thinning process described in these patents reduces the character to a single line configuration, the amount of binary data necessary to describe such thinned character is still substantial.
It is therefore an ob~ect of this invention to provide 8 system which digitally compresses an image to the least amount of digital information necessary to describe the image.
It is another ob~ect of this invention to provide a system which reduces the character to either its left-hand or right-hand boundary shape.
, ~.
i - 2 -10855~4 It is a further ob~ect of this invention to provide a system which describes the compressed image in more than one characteristic and then selects that characteristic which requires the minimum number of bits to describe the compressed image.
Summa~ of the Invention In order to achieve the above-described ob~ects, the present invention discloses an apparatus and method which - - describes digitally the shape of the left-hand boundary of a character in the form of a sequence of Freeman direction codes.
The number of binary bits required to describe the thinned character in terms of the number of corners in the character, the number of bits in the total area of the character and the number of bits required to describe the character in the ! Freeman code is then determined, with the system then selecting that one of such descriptions which describes the character in the minimum number of binary bits. To further define the lnvention, there ig disclosed in a pattern recognition process including the steps of scanning a pattern and digitizing elemental areas of the pattern into pattern digital bits and background digital bits, a pattern description method compris-ing the steps of storing in a first storage means the pattern digital bits representing a character thinned through a boundary reduction technique, fonming vector representation of line segments connecting ad~acent pattern digital bit~ stored in said first storage means, determining the amount of digital bits ~n said vector representation, determining the amount of digital bits in said pattern digital bits and background B
digital bits, and selecting the minimum amount of digital bits determined in said vector representation and said pattern digital bits and backgraund digital bits for recording on a recording medium.
Brief Description of the Drawin~s Fig. 1 is a echematic block diagram of the system for compressing the data representing a character or any other informational design in accordance with the teaching of the present invention.
Fig. 2 is a diagram showing the binary representation of the vector directions for use in describing the boundary of the character.
Fig. 3 shows a representation of the boundary of the ,~
/
' /
: ~ /
., ~
. /
.
- 3a -:, '' ~
.
108551g character in the form of a string of vectors.
Fig. 4 shows a representation of the boundary of the character before the thinning process.
Fig. 5 shcws a repre~entation of the boundary of the char~cter after the thinning process.
Fig. 6 is a block diagram of the processor useful in con~unction with the sy~tem of Fig. 1.
Fig 7 is a diagram of the m~sk logic employed in the present invention.
Flg. 8 is a mask logic circult employed in the present invention.
Figs. 9A-9H, inclusive, illustrates in flow diagram fonm the operation of a portion of the system illustrated in Fig. 1.
Description of the Preferred Embodiment Referring to Fig. 1, there is shown a schematic representation of the document processing system embodying the present invention. A chaxacter 20 which is to be recorded is, by way of example, the alphabetic character "B" located on a document 22, the document moving past the scanning station by suitable document transport means, such ~ransport means bein~
not shown since ~t i8 not part of the present invention. It i~, of course, o~vious that other forms of characters or de~igns - found on a document can be recorded e~ually as well as that i~lustrated. The character 20 i8 scanned by a flying spot scanner of well known construction although it i9 obvious th~t other types of scanning systems may be employed ~n the present embodiment to provide the required data output. The flying spot scanner lnclude~ an optical system symbolically illustrated by a cathode ray tube 24 having ~can control~ 26 which produce a suitable scanning raster. The video data produced by the scanning of the character 20 is fonmed as a result of the re-flec~ed light being transmitted through a photomultiplier tube 28 or other type of photoresponsive device which supplies the video signals to a digltizer 30, the latter ope~ating to generate the cartesian axis coordinates of a binary bit "l'J for the dark area scanned by the scanner and a binary bit "O" for llght ~reas in a manner well known in the art. Tha result of ~his digit~ing operation is outputted over line 31 to a processing unit 32 for storage and image co~pression after which the minimum amount of digital data necessary to describe the character i8 transmitted from the processing unit 32 to a recording device 34 for recording on magnetic tape, disc or other form of recording medium.
In data compression or reduction as utilized in its present embodiment, all blank areas on the document are consid-ered unnecessary and are therefore deleted. The method of data reduction used in the present invention comprises di~itally describing the shape, position and size of each character using the least amount of binary data information. One method of de~cribing the character is ~he employment of a vector sequence which consists of a string of numerical digits describing the 10855~4 shape as a sequence of unit length steps. As disclosed more fully in the article "On the Encoding of Arbitrary Geametric Configurations", by H. Freeman, IRE Trans. Elect. Computers, Vol. EC-10, No. 2, 1961, Pages 260 to 268, the boundary of the character is represented by a sequence of small vectors each of unit length and a limited set of possible directions. As shown in Fig. 2, each vector generally indicated by the numeral 33 iQ assigned, for example, one of the numbers 1-8, inclusive, with each point on the boundary of the character being located 10 relative to the previous one by the coded vector digit indicat- -lng directlon of the slope of the edge of the character. When representing the coded vector dig~t in binary form, only 3 bits are required for each digit.
Using this method of vector direction, the character 20 can be represented as a string of digits (Fig. 3) which lndicates the direction of the slope of each succeeding vector unit. Thus, a horizontal line portion of the character extend-in8 to the right would be represented by a series of one's while a vertical line portion extending in an upright direction would be represented by a series of seven's. It is further obvious th~t a change in the vector string from one ve~tor dlgit to another vector digit would ~ignify a corner in the boundary of the character and the direction of the slope leading into and ~` out of the corner. The present method of data compression includes tracing the outer boundary of the character 20 (Fig. 4) utilizing the binary digit one's, each of wh~ h forms the points ~`
:.~
in the boundary of the character. As each point in the boundary i8 traced by the system, the vector direction between the point being traced and the last point traced is developed and s~ored.
The system also determines the thickness of the chsracter at each point in the boundary. At the end of the first boundary tracing operation, data describlng the outer boundary of the character in tenms of a series of vectors representing the direction between each succeeding point in the boundary will have been developed. This character is ~hown in Fig. 3. A
second tracing operstion of the character 20 is then initiated and utilizing the value of the thickness generated in the first tracing operstion, a reduced or thinned character i8 developed using the left-hant boundary as the form of the character (Fig.
5). Then uslng the thinned character a8 the model, data i8 developed which ~ay be used in describing the thinned character in several different way~ with the description requiring the least amount of binary data being selected for use in recordin~
the character on a recording medLum.
Referring now to Fig. 6, there is shown in block form the ba~ic components of the processing unit 32 (Fig. l) which ; are used to reduce the binary data representing the character 20 to a mlnimum value. Included in the processing unit is a logic unit 35 for determining the vector string of the charac-ter, a fir~t ~torsge unit 36 for storing blnary data bits in an image representing the shape of the character scanned. Each data b~t is stored in a location in the storge unit 36 corres-1085~;~4 ponding to the cartesian coordinates of the zero's and one's which comprise the light and dark areas of the document as shown in Fig. 3, which data is received from the digitize~ 30 (Fig. 1) over line 31. A second storage unit 37 i8 provided for storing the binary data representing the character in the thinned condition.
The processing unit 32 further includes a storage unit 38 for storing the data representing the boundary of the character in terms of vectors, includes a storage unit 39 for .~ 10 storing data representing the coordinate of the upper limits of the character, which coordinate will represent the height of : the character after the tracing of the character h~s occurred, include~ a storage unit 40 which functions to store the width of the characters ln the same manner as the height a8 determined ln storage unit 39, includes a storage unit 41 which stores data representing the thickness of the character, which data i8 used in producing a thinned configuration of the character in terms '~ of the left-hand boundary of the charscter, lnclude~ a stor~ge .. l unit 29 for storing the coordinate of the corners in the left-, 20 hand boundary of the character, includes a storage unit 42 for .1 stor~ng tata representing the total number of bits on the picture of the thinned character detenmined by multiplying the `~ height and width of the character utilizing the data in storage units 39 and 40, includes a storage unit 43 for storing data representing the total number of bits in describing the thinned ~,, characters in tenms of a vector sequence, and includes a storage ' . .
:: - 8 -unit 44 for storing data representing the total number of bits in describing the reduced or thinned character in tenms of the corners in the character. The data stored in the storage units ~2, 43 and 44 are used in selecting which characteristic of the reduced or th~nned character will be used in describing the character when the data representlng the character i8 recorded on a recording medium.
The processing unit 32 further iw lude~ an arithmetic unit 46, performing multiplication, subtraction, etc. opera-10 ~lon~ for use in determining the total number of bits in the character picture, a minimum binary comparator unit 47 for comparing the data in the storage units 42-44, inclusive, to ~elect the minimNm number of bits in the storage units 42-44, inclusive, and a plurality of counters 48, 49, 50 ant 51 for U88 in developing the data stored in storage units 42, 43 and 44. Included in the unit 32 are a plurality of the representa-ti~ely illu~trated thickness counters 48 and 49, esch represent-ing a predetermined thickness of each leg of the character, which coNnters 48 and 49 count the number of times the corres-20 ponding thickness is detenmined during the trscing of the outer I boundary of the character. In this regard, counter 48 will I count the unit thickness while counter 49 will count tho~e . .
thicknesses having 8 width of "n" unit, "n" being any predeter-, J mined integer. When considering an E13-B character font, from two to twenty of such counters 48 and 49 would normally be desired. As each point in the outer boundary of the character ;' _ g _ 10855~4 is sensed by the system during the tracing of the character, the system will determine the thickness of the character at that point and increment the appropriate one of the overall counters 48 and 49 which corresponds to the thickness found. At the end of the boundary tracing of the character, the thicknesQ counter hsving the hi8hest number of count on its output will represent the thickness of the character. Thi8 data representing the thickness is then stored in storage unit 41. Counter 50 wlll count the number of points of the boundary of the thinned character while counter 51 will count the number of vectors ~n the boundary of the thinned character. Counter S0 is used to check that the character i8 traced completely while counter 51 is used in determining the amount of data requred to describe the character u~ing vectors as the means of description.
Referring now to Figs. 7 and 8, there i8 shown the mask logic used in detenmining the vector direction between succeeding points in the boundary of the character. As fully disclosed in the previously-cited articles by H. Freeman, the mask logic shown in Fig. 7 is applied to each point in the boundary of the character 20. Depending on which of the aress i Pl-P8, inclusive, of the mask logic a blnary digit one 18 located will determine the coded digit representation of the vector extending between the point on which the mask logic is located and which i8 positioned in the center area of the mask logic shown in Fig. 7 and the point located in one of the areas Pl-P8, inclusive. Thus, if a binary one is found in area P8, the .
vector direction i~ eight.
Fig. 8 shows a logic circuit which may be used in determining the vector digit between succeeding points in the boundary of the character in accordance with the mask logic of Fig. 7. Included in the circuit is a counter 52 having eight outputs 53, each corresponding to one of the vector directions ~3 (Fig. 2). Each output 53 is connected to an input of an assoclated AND gate 54 whose other input i8 connected over line 55 to the storage unit 36 (Fig. 6) ln which the image of the : 10 character 20 (Fig. 1) is stored. The output 56 of each of the AND gates 54 is connected to sn OR gate 57 and to a decoder 58 whose output is connected to the vector string storge un$t 38 (Fig. 6). The counter 52 $8 of the type wh$ch counts in the :, directlon of the arrow shown in Fig. 8 and is preloaded over llne 59 from the program control unit 45 (F~g. 6) to start counting at any of the outputs 53, such counter 52 operat$ng by clock pulses transm$tted over line 60 from the control unit ', 45 in a manner well known in the art. Upon sensing a binary one in the ima8e (Fig. 4) of the character 20 (Fig. 1) stored in the storage unlt 36 (Fig. 6), the system will output over line 5S
the binary digit in each of the areas Pl-P8, inclusive, of the . mask logic (Fig. 7) which correspond to the areas ad~acent the point in the ~ma8e. Depending on the boundary portion of the char~cter being traced, as will be described more fully here~n-` after, the control unit 45 will preload the counter 52 to ~elect which output of the counter will provide a binary one to its ~, iO85514 associated AND gate 54 over line 53. Thi~ will enable the AND
gate 54 if its other input 55 has a binary one located therein, which occurred as a result of a binary one being found in the corresponding area of the mask (Fig. 7), and which binary one was outputted to the AN~ Bate 54 from the storage unit 36. As mentioned above, the output of the AND gate S4 i8 connected to the OR gate 57 and to a decoder 58 which decodes the ~ignal to itentify the direction of the vector. The output of the decoder 58 is connected to the storage unit 38 (Fig. 6) for storing the digit representing the vector direction. ~he output of the OR
gate 57 i8 connected to the program control unit 45 to signal such control unit to proceed with the next operation. As fully d~sclosed in the above Freeman article, the mask logic (Fig. 7) i8 rotated 80 that direction P7 of the mask log~c is always orientated ~ a direction parallel with the vecbor digit found when the mask is moved to the next point in order to determine the next vector.
Referring now to Figs. 9A-9H, inclusive, there is shown in flow chart form the method used by the system in deter-mining the minimNm amount of binary data blts required todescribe the character 20 (Fig. 1). The system (Fig. 1) is $nitiated by the scanning of the document 22 starting, for example, at the lower left-hand corner, and ~canning in a hori-zontal direction one line at a time until the entire document ha8 been scanned, the video signals being digitized in the digltizer 30 to provite an image of the document which is stored (block 66 of Fig. 9A) in the image storage unit 36 (Fig. 6) of the processing un~t 32, which storage unit 36 may be a portion of the memory of a microprocessor or other similar type of data processor unit. The data from the digitizer 30 de~cribing each point of the image in the sborage unit 36 includes the cartesion coordinates representing the location of the point in the orig-inal document together with the binary bit in that location.
After the image of the document 22 has been stored ~n the storage unit 36 (Fig..6) in the form of binary onels snd zero's as shown in Fig. 4, the system will initiate a ~earch of the character in the $mage stored in the storage unit 36. Start-ing at the lower left-hand corner of the image and moving hori-~ontally the system w$11 scan the array horizontally one point at a time (block 67 of Fig. 9A) looking for a 0 to 1 trans$tion.
The proces8 i8 repeated until a 0 to 1 trsnsition i8 sensed (block 68) indicating the finding of an edge of the character.
Once the ed~e of the character has been found, the coordinates of this first point is stored in storage units 39 and 40 (Fig. 6) and a tracing of the outer boundar~ of the character is initlat-ed (block 69) to define the boundary of the character a8 a strlngof binary dlgits, the digits representing the vectors belng 9tored in the storage unit 38 (Fig. 6). The operation of the mask logic (Fig. 8) is initiated (block 69) by sensing first in direction 6 (Fig. 2), then moving in a clockwise direct$on to direction 7, tirection 8, etc. until it sen~es a binary one (block 70) which i8 the next point in the boundary of the char-acter. After finding the next point, the digit representing the direction of the vector is stored (block 70 of Fig. 9A) in storage unit 38 (Fig. 6). The system will move (bloc~ 72) to the point found and start searching until the next binary one i8 found. The mask logic (Fig. 8) will always initiate the ~ensing operation in a direction at 90 counter-clockwise to the ~ector direction found at the last point under the control of the control unit 45 (Fig. 6). Referring to Fi8. 3, it will be seen that as 8 result of this first tracing operation, the boundary of the character 20 can be defined as a series of binary digits representing a string of vector d~rections stored in the storage unit 38 (Fig. 6).
After the system moves to the next point found, the system will check the vector direction for vector directions 1 and 5 (block 74) which, if found, indicates that the boundary of the character i8 moving in a horizontal direction. In the case of the character 20 shown in Fig. 4, the top edge will be traced as a result of this operation. When the system deter-mines that the direction of the vector i8 one or five, it will detenmine (block 76 of Fig. 9B) if the coordinates of the point are new llmits in the X or Y direction. If no new limits are f~und, the next point (block 70 of Fig. 9A) i8 then searchet.
.1 , The coordinates of those points found which are new limits (block 76 of Fi8. 9B) are stored (block 78) in storage unit~
39 and 40 (Fig. 6) with the coordinate~ of the character in the i X direction being stored in storage unit 40 and the coordinate in the Y direction being stored in storage unit 39 (Fig. 6).
The coordinates of each point sen~ed will be compared with the coordinates previously stored to determine lf the coordinates are a new limit in the X and Y direction. At the end of the tracing operation the storage units 3g and 40 will contain the maximum coordinates~ of the character in the X ~nd Y direction ~rom which the height and width of the character can be determ1ned. The system will slso compare (block 80-Fig. 9B) the coordlnates of the point sensed with the coordinates of the first point sensd which, if there is a comparison would indicate that the tracing of the boundary of the characters has reached the orlginal ~tarting point. If the coordinates are different, the next point (block 70) i8 searched.
If the vector direction i8 found (block 74) not ~o be in the 1 or 5 direction, the vector direction is tested for dlrectlons 2, 3 or 4 (block 82 of Fig. 9B). If the vector i8 found to correspond to one of these directions indicat~ng that the boundary ~ the character i8 extending in a downward direc-tion, the system will search in a direction 90 to the vector direction of the point sensed and in the direction 5 (F~g. 2) thereby mea~uring (block 84-Fig. 9B) the thickness of the character at this point. In meesuring the thickness, the system will count the number of binary one's sensed before finding a binary one to zero transition indicating the end of the portion of the character being sensed. As disclosed previously, located in the processor unit 32 (Fig. 6) are a plurality of counters 10855~4 represented by counters 48 and 49, each representing a predeter-mined character thickness. Thus, there may be a counter for each thickness of between 2 and 15 where the thickness i~ in terms of unit length. Each counter will count the number of times the thickness i8 found during the tracing of the character. The counters 48-49 corresponding to the thickness determined by the system will then be incremented by one (block 86) and the point will be tested (block 76) to detersnine if the coordinates of the point set a new limit.
If the result of block 82 is negative, the direction of the vectors is tested in direction~ 6, 7 or 8 (block 88) with the thickness being measured at r~ght angles to the vector direction and in direction 1 (block 90 of Fig. 9B) with the appropriate thickness counters48-49 being incremented (block 86) by one. This tracing of the boundary of the character 20 and the thickness measurement will continue until the last point in the boundary that i8 sensed has its coordinates corresponding to the first point ~easured (block 80) indicating that the traclng operation i8 complete. The thickness of the character is then 20 selected (block 92 of Fig. 96) by taking the highest count found in the thickness counters 48-49, inclusive, and storing the re8ult in the storage unit 41 (Fig. 6) A second boundary tracing of the lmage (Fig. 3) in storage unit 36 (Fig. 6) i8 then initiated (block 96) to reduce the boundary of the character 20 by replacing the character image shawn in Fig. 4 with the character image shwm in Fig. 5, which is the "left-hand" boundary of the character 20 (Flg. 1).
During thi~ second tracing of the character image, the thick-ness determined in block 92 and now stored in ~torage unit 41 (Fig. 6) w~ll be used to place each point that i~ found when moving in directions 2, 3 or 4 at a point in the storage unit 37 which will correspond to the left-hand boundary of the character 20. This point will be moved in direction 5 to the new location in the storage unit 37 which is equal to the thickness of the character. During this second tracing of the outer boundary of the image, the vector direction found i8 tested to detenmine in which direction the slope of the edge of the character 20 i8 proceeding ~o as to apply the above boundary conditions when the vectors indicate the slope to be in directions 2, 3 or 4.
The tracing operation i8 initiated (block 96) in the same manner as was previously described with respect to the first tracing operation with the operation of blocks 98, 100 and 102 of Fig.
9C perfonming the same function as blocks 70, 71 and 72, respectfully of Fig. 9A. The vector d~rection i8 tested (block 104) for directions 6, 7, or 8 to detenmine if the left-hand boundary of the thinned character image (Fig. 5) is being traced.
If it is found true, the coordinates of the point found are used to check (block 106) the same coordinates in the storage unit 37 to detenmine if a binary one has already been inserted in that location. If a binary one is found in that location, the system will start a new point search (block 98 of Fig. 9C) after going through block 128 of Fig. 9E testing for the start-ing point. This procedure insures that the thinnin8 operatlon does not create a new point location in the storage unit 37 that was not in the original image in storage unit 36.
If block 106 finds no point at that location in the picture in storage unit 37, the boundary point cuunter S0 (Fig.
which count~ the number of points in the boundary of the character i~ incremented by one (block 108 of Fig. 9D) and a binary one is then inserted (block 110) in the storage unit 37 at the same coordinates. The next point (block 98 of Fig. 9C) 0 i8 then found after going through block 128 of Fig. 9E. The point counter 50 is used in a succeeding tracing operation to insure that the boundary of the character i8 completely traced a~ will be explained more fully hereinafter.
If block 104 of Fig. 9C finds that the vector direc-tlon is not in directions 6, 7 or 8, the system will then move the coordinates of the point in the 5 direction (block 112) a distance equal to the thickness to determine ~f there i8 a one in that location in the image in storage unit 36. Thi8 procedure will check each point of the image for an edge point.
If a ~ne i~ not located, as a result of the operation of block 112, the thickness i~ reduced by one unit (block 114~of Fig. 9D) ~ntil the thickness ~s ~ero (block 116) indicating the point i8 an edge point. If a binary one i8 found when the thickness is not zero (block 116) indicating the point is not a part of the character, the system will search for the next point (block 90 of Fig. 9C). The direction to the next point i8 tested for continuation in the one direction (block 118 of Fig. 9D) or continuation in the five direction (block 120), both operations testing for the horizontal edge in the upper portion (block 118) and the lower portion (block 122) of the image. By identifying the horizontal edges of the character, the binary data requ1red to describe the character can be reduced by indicating the number of times the vector digit is repeated. This procedure also determines that portion of the outer boundary of the character which will be reduced by the thlckness of the character to establish the left-hand boundary of the character as will now be described.
If blocks 118 and 120 indicate that the direction to the next point under consideration is neither in direction five or one (Fig. 2), the point must be located in a vector directlon of either 2, 3, or 4 estsblishing that the right-hand boundary of the character 20 (Fig, 4) i8 being traced requiring that the point be positioned in the left-hand boundary of the picture in the storflge unlt 37. Block 122 tests the left-hand boundary po~ition of the point in the storage unit 37 by moving the coordinates of the point in the five direction equal to the value of the thickness stored in storage unit 41 (Fig. 6) to see if a one is already in that position. If it is, block 128 is tested and lf it is not the starting point, the next point 18 traced (block 98). If a one is not found in the tested position, the point counter 50 is incremented by one (block 124), the coordin-ates of the left-hand boundary points are ~tored in the width storage unit 40 (block 125) as the new width limit~, and a binary one i5 placed in the storage unit 37 (block 126-Fig 9E) corresponding to the left-hand boundary of the image in storage unit 36. This procedure i8 repeated until the coordinates of the last point traced coincide with the coordinstes of the starting point (block 128) indicating that the boundary (block 130) of theimage in the storage unit 36 has been traced.
The picture of the thinned character i8 then transferred to the ~torage unit 36 (block 132) replacing the origlnRl image. The bit counter 50 at this time will contain a count equal to the i number of points in the boundary of the thinned character which will be used to check the tracing of the boundary in a manner that will now be described.
The thinned character (Fig. 5) now stored in storage unlt 36 i8 te~ted to determine the number of bits required to describe the character when using the vector string as the descriptive method. The left-hand boundary tracing procedure of the thinned character i8 initiated (block 134 of Fig. 9E) and the first point i8 checked to determine if its coordinate~
are that of the starting point (block 136). The bit counter 50 i8 also checked to determine if its output i8 zero which would indicate th~t a boundary tracing of the thinned character has already taken place. This eliminates a problem found in tracing a character such as a "~' where the starting point may be reached before completing the tracing of the character. If the coordinate~ do not correspond to the ~tarting point or if 10855~4 counter ~0 is not zero (block 137) the next point is found ~-(block 138 of Fig. 9F). Block 140 detenmine6 the vector direction to the next point, the system moving to thst point (block 142). The binary digit representing the vec~or ~8 then ~-stored in storage unit 38, the vector counter 51 (Fig. 6) i8 incremented by one (block 146), the point counter 50 i8 decremented by one (block 147), and the vector counter 51 is checked (block 148) to determine if the count at this t~me i8 less than a predetermined number n, which in the present example i8 256, n being the capAcity of the storage unit 38.
If the count in the counter 51 is less than n, the sequence i8 rrepeated until the thinned character has been traced and the , storage unit 38 will contain binary data representing the ', vectors in the left-hand boundary of the thlnned character.
The vector number counter 51 wlll contain a count equal to , the number of vectors ln the thinned character (Fig. S) at thl~
time.
Block 150 will inltiate a trace of the vector string , stored in storage unit 38 to detenmine the number of corners '~ 20 in the thinned chsracter. As previously described, each change of vector direction between two ~uccessive vect3rs in the vector string is a corner and a character can be described by knowing the cartesian coordinates of each corner listed in the sequence that they are found. Block 152 of Fig. 9G will compare each 3 vector with respect to the last to determine if a corner exists.
; A corner,exlsts when there i5 no agreement between two successive "
' ~ O 8 ~ ~ 4 vectors~ If there i8 agreement, the vectors will continue (block 150) to be compared until a corner i~ found. Once found, the location of the corner in tenms of its cartesian coordinates (block 154), is stored (block 156) in ~torage unit 29 (Fig. 6) and the corner coordinate~ are cumpared with the first corner (block 158) to find the starting point. If the coordinates are not the starting point, the next two vectors are compared (block 150) to find the next corner. This process continues until the starting point is reached (block 158).
At this time the number of binary bits representing the corners stored in storage unit 29 is determined (block 160) utilizing the arithmetic unit 46 (Fig 6) in a manner well known in the art with the result stored in the storage unit 44.
Additionally, the count number in the counter 51 is multiplied by three in the arithmet~c unit 46 to produce the number of bit8 required to define the characters ln terms offthe vector string. This result i8 stored in storage unit 43. The number ; of blnary bits in the picture of the thinned character is determined ~block 164) by mult~plying the height ~imes the width using the limits stored in storage units 39, 40 durlng the tracing of the charac~er ~n block 78.and in block 125.
Thi~ result is stored in storage unit 42 (Fig. 6). The outputs of the data stored in storage units 29, 42 and 43 are then outputted to a minimum binary comparator 47 (Fig. 6) to compare and select (blocks 166, 168, and 170 of Fig. 9H) which of the storage unitscontains the m~nimNm number of binary bits. The comparator 47 will compare two of the values in storage units 29, 42 and 43 determining the minimum of the two and then compare this result with the third value. The binary data selected i8 then outputted (blocks 172, 174 and 176) o~er line 177 to a recording unit 34 (Fig. 1) for recording the selected character representing data in a recording medium for further processing. The processing-,of the document 22 (Fig. 1) in accordance with the present invention i~ then continued until completed (block 178).
While the invention has been shown and described in tenms of a preferred embodiment thereof, it will be understood that this invention i8 not limited to this particular embod~ment and that many changes and modifications may be made without departing fro~ the true spirit and scope of the invention as defined in the appended claims. It is therefore obvious that the right-hand boundary of the character may be developed for u~e in determining the minimum amount of data which describes the character.
It is therefore an ob~ect of this invention to provide 8 system which digitally compresses an image to the least amount of digital information necessary to describe the image.
It is another ob~ect of this invention to provide a system which reduces the character to either its left-hand or right-hand boundary shape.
, ~.
i - 2 -10855~4 It is a further ob~ect of this invention to provide a system which describes the compressed image in more than one characteristic and then selects that characteristic which requires the minimum number of bits to describe the compressed image.
Summa~ of the Invention In order to achieve the above-described ob~ects, the present invention discloses an apparatus and method which - - describes digitally the shape of the left-hand boundary of a character in the form of a sequence of Freeman direction codes.
The number of binary bits required to describe the thinned character in terms of the number of corners in the character, the number of bits in the total area of the character and the number of bits required to describe the character in the ! Freeman code is then determined, with the system then selecting that one of such descriptions which describes the character in the minimum number of binary bits. To further define the lnvention, there ig disclosed in a pattern recognition process including the steps of scanning a pattern and digitizing elemental areas of the pattern into pattern digital bits and background digital bits, a pattern description method compris-ing the steps of storing in a first storage means the pattern digital bits representing a character thinned through a boundary reduction technique, fonming vector representation of line segments connecting ad~acent pattern digital bit~ stored in said first storage means, determining the amount of digital bits ~n said vector representation, determining the amount of digital bits in said pattern digital bits and background B
digital bits, and selecting the minimum amount of digital bits determined in said vector representation and said pattern digital bits and backgraund digital bits for recording on a recording medium.
Brief Description of the Drawin~s Fig. 1 is a echematic block diagram of the system for compressing the data representing a character or any other informational design in accordance with the teaching of the present invention.
Fig. 2 is a diagram showing the binary representation of the vector directions for use in describing the boundary of the character.
Fig. 3 shows a representation of the boundary of the ,~
/
' /
: ~ /
., ~
. /
.
- 3a -:, '' ~
.
108551g character in the form of a string of vectors.
Fig. 4 shows a representation of the boundary of the character before the thinning process.
Fig. 5 shcws a repre~entation of the boundary of the char~cter after the thinning process.
Fig. 6 is a block diagram of the processor useful in con~unction with the sy~tem of Fig. 1.
Fig 7 is a diagram of the m~sk logic employed in the present invention.
Flg. 8 is a mask logic circult employed in the present invention.
Figs. 9A-9H, inclusive, illustrates in flow diagram fonm the operation of a portion of the system illustrated in Fig. 1.
Description of the Preferred Embodiment Referring to Fig. 1, there is shown a schematic representation of the document processing system embodying the present invention. A chaxacter 20 which is to be recorded is, by way of example, the alphabetic character "B" located on a document 22, the document moving past the scanning station by suitable document transport means, such ~ransport means bein~
not shown since ~t i8 not part of the present invention. It i~, of course, o~vious that other forms of characters or de~igns - found on a document can be recorded e~ually as well as that i~lustrated. The character 20 i8 scanned by a flying spot scanner of well known construction although it i9 obvious th~t other types of scanning systems may be employed ~n the present embodiment to provide the required data output. The flying spot scanner lnclude~ an optical system symbolically illustrated by a cathode ray tube 24 having ~can control~ 26 which produce a suitable scanning raster. The video data produced by the scanning of the character 20 is fonmed as a result of the re-flec~ed light being transmitted through a photomultiplier tube 28 or other type of photoresponsive device which supplies the video signals to a digltizer 30, the latter ope~ating to generate the cartesian axis coordinates of a binary bit "l'J for the dark area scanned by the scanner and a binary bit "O" for llght ~reas in a manner well known in the art. Tha result of ~his digit~ing operation is outputted over line 31 to a processing unit 32 for storage and image co~pression after which the minimum amount of digital data necessary to describe the character i8 transmitted from the processing unit 32 to a recording device 34 for recording on magnetic tape, disc or other form of recording medium.
In data compression or reduction as utilized in its present embodiment, all blank areas on the document are consid-ered unnecessary and are therefore deleted. The method of data reduction used in the present invention comprises di~itally describing the shape, position and size of each character using the least amount of binary data information. One method of de~cribing the character is ~he employment of a vector sequence which consists of a string of numerical digits describing the 10855~4 shape as a sequence of unit length steps. As disclosed more fully in the article "On the Encoding of Arbitrary Geametric Configurations", by H. Freeman, IRE Trans. Elect. Computers, Vol. EC-10, No. 2, 1961, Pages 260 to 268, the boundary of the character is represented by a sequence of small vectors each of unit length and a limited set of possible directions. As shown in Fig. 2, each vector generally indicated by the numeral 33 iQ assigned, for example, one of the numbers 1-8, inclusive, with each point on the boundary of the character being located 10 relative to the previous one by the coded vector digit indicat- -lng directlon of the slope of the edge of the character. When representing the coded vector dig~t in binary form, only 3 bits are required for each digit.
Using this method of vector direction, the character 20 can be represented as a string of digits (Fig. 3) which lndicates the direction of the slope of each succeeding vector unit. Thus, a horizontal line portion of the character extend-in8 to the right would be represented by a series of one's while a vertical line portion extending in an upright direction would be represented by a series of seven's. It is further obvious th~t a change in the vector string from one ve~tor dlgit to another vector digit would ~ignify a corner in the boundary of the character and the direction of the slope leading into and ~` out of the corner. The present method of data compression includes tracing the outer boundary of the character 20 (Fig. 4) utilizing the binary digit one's, each of wh~ h forms the points ~`
:.~
in the boundary of the character. As each point in the boundary i8 traced by the system, the vector direction between the point being traced and the last point traced is developed and s~ored.
The system also determines the thickness of the chsracter at each point in the boundary. At the end of the first boundary tracing operation, data describlng the outer boundary of the character in tenms of a series of vectors representing the direction between each succeeding point in the boundary will have been developed. This character is ~hown in Fig. 3. A
second tracing operstion of the character 20 is then initiated and utilizing the value of the thickness generated in the first tracing operstion, a reduced or thinned character i8 developed using the left-hant boundary as the form of the character (Fig.
5). Then uslng the thinned character a8 the model, data i8 developed which ~ay be used in describing the thinned character in several different way~ with the description requiring the least amount of binary data being selected for use in recordin~
the character on a recording medLum.
Referring now to Fig. 6, there is shown in block form the ba~ic components of the processing unit 32 (Fig. l) which ; are used to reduce the binary data representing the character 20 to a mlnimum value. Included in the processing unit is a logic unit 35 for determining the vector string of the charac-ter, a fir~t ~torsge unit 36 for storing blnary data bits in an image representing the shape of the character scanned. Each data b~t is stored in a location in the storge unit 36 corres-1085~;~4 ponding to the cartesian coordinates of the zero's and one's which comprise the light and dark areas of the document as shown in Fig. 3, which data is received from the digitize~ 30 (Fig. 1) over line 31. A second storage unit 37 i8 provided for storing the binary data representing the character in the thinned condition.
The processing unit 32 further includes a storage unit 38 for storing the data representing the boundary of the character in terms of vectors, includes a storage unit 39 for .~ 10 storing data representing the coordinate of the upper limits of the character, which coordinate will represent the height of : the character after the tracing of the character h~s occurred, include~ a storage unit 40 which functions to store the width of the characters ln the same manner as the height a8 determined ln storage unit 39, includes a storage unit 41 which stores data representing the thickness of the character, which data i8 used in producing a thinned configuration of the character in terms '~ of the left-hand boundary of the charscter, lnclude~ a stor~ge .. l unit 29 for storing the coordinate of the corners in the left-, 20 hand boundary of the character, includes a storage unit 42 for .1 stor~ng tata representing the total number of bits on the picture of the thinned character detenmined by multiplying the `~ height and width of the character utilizing the data in storage units 39 and 40, includes a storage unit 43 for storing data representing the total number of bits in describing the thinned ~,, characters in tenms of a vector sequence, and includes a storage ' . .
:: - 8 -unit 44 for storing data representing the total number of bits in describing the reduced or thinned character in tenms of the corners in the character. The data stored in the storage units ~2, 43 and 44 are used in selecting which characteristic of the reduced or th~nned character will be used in describing the character when the data representlng the character i8 recorded on a recording medium.
The processing unit 32 further iw lude~ an arithmetic unit 46, performing multiplication, subtraction, etc. opera-10 ~lon~ for use in determining the total number of bits in the character picture, a minimum binary comparator unit 47 for comparing the data in the storage units 42-44, inclusive, to ~elect the minimNm number of bits in the storage units 42-44, inclusive, and a plurality of counters 48, 49, 50 ant 51 for U88 in developing the data stored in storage units 42, 43 and 44. Included in the unit 32 are a plurality of the representa-ti~ely illu~trated thickness counters 48 and 49, esch represent-ing a predetermined thickness of each leg of the character, which coNnters 48 and 49 count the number of times the corres-20 ponding thickness is detenmined during the trscing of the outer I boundary of the character. In this regard, counter 48 will I count the unit thickness while counter 49 will count tho~e . .
thicknesses having 8 width of "n" unit, "n" being any predeter-, J mined integer. When considering an E13-B character font, from two to twenty of such counters 48 and 49 would normally be desired. As each point in the outer boundary of the character ;' _ g _ 10855~4 is sensed by the system during the tracing of the character, the system will determine the thickness of the character at that point and increment the appropriate one of the overall counters 48 and 49 which corresponds to the thickness found. At the end of the boundary tracing of the character, the thicknesQ counter hsving the hi8hest number of count on its output will represent the thickness of the character. Thi8 data representing the thickness is then stored in storage unit 41. Counter 50 wlll count the number of points of the boundary of the thinned character while counter 51 will count the number of vectors ~n the boundary of the thinned character. Counter S0 is used to check that the character i8 traced completely while counter 51 is used in determining the amount of data requred to describe the character u~ing vectors as the means of description.
Referring now to Figs. 7 and 8, there i8 shown the mask logic used in detenmining the vector direction between succeeding points in the boundary of the character. As fully disclosed in the previously-cited articles by H. Freeman, the mask logic shown in Fig. 7 is applied to each point in the boundary of the character 20. Depending on which of the aress i Pl-P8, inclusive, of the mask logic a blnary digit one 18 located will determine the coded digit representation of the vector extending between the point on which the mask logic is located and which i8 positioned in the center area of the mask logic shown in Fig. 7 and the point located in one of the areas Pl-P8, inclusive. Thus, if a binary one is found in area P8, the .
vector direction i~ eight.
Fig. 8 shows a logic circuit which may be used in determining the vector digit between succeeding points in the boundary of the character in accordance with the mask logic of Fig. 7. Included in the circuit is a counter 52 having eight outputs 53, each corresponding to one of the vector directions ~3 (Fig. 2). Each output 53 is connected to an input of an assoclated AND gate 54 whose other input i8 connected over line 55 to the storage unit 36 (Fig. 6) ln which the image of the : 10 character 20 (Fig. 1) is stored. The output 56 of each of the AND gates 54 is connected to sn OR gate 57 and to a decoder 58 whose output is connected to the vector string storge un$t 38 (Fig. 6). The counter 52 $8 of the type wh$ch counts in the :, directlon of the arrow shown in Fig. 8 and is preloaded over llne 59 from the program control unit 45 (F~g. 6) to start counting at any of the outputs 53, such counter 52 operat$ng by clock pulses transm$tted over line 60 from the control unit ', 45 in a manner well known in the art. Upon sensing a binary one in the ima8e (Fig. 4) of the character 20 (Fig. 1) stored in the storage unlt 36 (Fig. 6), the system will output over line 5S
the binary digit in each of the areas Pl-P8, inclusive, of the . mask logic (Fig. 7) which correspond to the areas ad~acent the point in the ~ma8e. Depending on the boundary portion of the char~cter being traced, as will be described more fully here~n-` after, the control unit 45 will preload the counter 52 to ~elect which output of the counter will provide a binary one to its ~, iO85514 associated AND gate 54 over line 53. Thi~ will enable the AND
gate 54 if its other input 55 has a binary one located therein, which occurred as a result of a binary one being found in the corresponding area of the mask (Fig. 7), and which binary one was outputted to the AN~ Bate 54 from the storage unit 36. As mentioned above, the output of the AND gate S4 i8 connected to the OR gate 57 and to a decoder 58 which decodes the ~ignal to itentify the direction of the vector. The output of the decoder 58 is connected to the storage unit 38 (Fig. 6) for storing the digit representing the vector direction. ~he output of the OR
gate 57 i8 connected to the program control unit 45 to signal such control unit to proceed with the next operation. As fully d~sclosed in the above Freeman article, the mask logic (Fig. 7) i8 rotated 80 that direction P7 of the mask log~c is always orientated ~ a direction parallel with the vecbor digit found when the mask is moved to the next point in order to determine the next vector.
Referring now to Figs. 9A-9H, inclusive, there is shown in flow chart form the method used by the system in deter-mining the minimNm amount of binary data blts required todescribe the character 20 (Fig. 1). The system (Fig. 1) is $nitiated by the scanning of the document 22 starting, for example, at the lower left-hand corner, and ~canning in a hori-zontal direction one line at a time until the entire document ha8 been scanned, the video signals being digitized in the digltizer 30 to provite an image of the document which is stored (block 66 of Fig. 9A) in the image storage unit 36 (Fig. 6) of the processing un~t 32, which storage unit 36 may be a portion of the memory of a microprocessor or other similar type of data processor unit. The data from the digitizer 30 de~cribing each point of the image in the sborage unit 36 includes the cartesion coordinates representing the location of the point in the orig-inal document together with the binary bit in that location.
After the image of the document 22 has been stored ~n the storage unit 36 (Fig..6) in the form of binary onels snd zero's as shown in Fig. 4, the system will initiate a ~earch of the character in the $mage stored in the storage unit 36. Start-ing at the lower left-hand corner of the image and moving hori-~ontally the system w$11 scan the array horizontally one point at a time (block 67 of Fig. 9A) looking for a 0 to 1 trans$tion.
The proces8 i8 repeated until a 0 to 1 trsnsition i8 sensed (block 68) indicating the finding of an edge of the character.
Once the ed~e of the character has been found, the coordinates of this first point is stored in storage units 39 and 40 (Fig. 6) and a tracing of the outer boundar~ of the character is initlat-ed (block 69) to define the boundary of the character a8 a strlngof binary dlgits, the digits representing the vectors belng 9tored in the storage unit 38 (Fig. 6). The operation of the mask logic (Fig. 8) is initiated (block 69) by sensing first in direction 6 (Fig. 2), then moving in a clockwise direct$on to direction 7, tirection 8, etc. until it sen~es a binary one (block 70) which i8 the next point in the boundary of the char-acter. After finding the next point, the digit representing the direction of the vector is stored (block 70 of Fig. 9A) in storage unit 38 (Fig. 6). The system will move (bloc~ 72) to the point found and start searching until the next binary one i8 found. The mask logic (Fig. 8) will always initiate the ~ensing operation in a direction at 90 counter-clockwise to the ~ector direction found at the last point under the control of the control unit 45 (Fig. 6). Referring to Fi8. 3, it will be seen that as 8 result of this first tracing operation, the boundary of the character 20 can be defined as a series of binary digits representing a string of vector d~rections stored in the storage unit 38 (Fig. 6).
After the system moves to the next point found, the system will check the vector direction for vector directions 1 and 5 (block 74) which, if found, indicates that the boundary of the character i8 moving in a horizontal direction. In the case of the character 20 shown in Fig. 4, the top edge will be traced as a result of this operation. When the system deter-mines that the direction of the vector i8 one or five, it will detenmine (block 76 of Fig. 9B) if the coordinates of the point are new llmits in the X or Y direction. If no new limits are f~und, the next point (block 70 of Fig. 9A) i8 then searchet.
.1 , The coordinates of those points found which are new limits (block 76 of Fi8. 9B) are stored (block 78) in storage unit~
39 and 40 (Fig. 6) with the coordinate~ of the character in the i X direction being stored in storage unit 40 and the coordinate in the Y direction being stored in storage unit 39 (Fig. 6).
The coordinates of each point sen~ed will be compared with the coordinates previously stored to determine lf the coordinates are a new limit in the X and Y direction. At the end of the tracing operation the storage units 3g and 40 will contain the maximum coordinates~ of the character in the X ~nd Y direction ~rom which the height and width of the character can be determ1ned. The system will slso compare (block 80-Fig. 9B) the coordlnates of the point sensed with the coordinates of the first point sensd which, if there is a comparison would indicate that the tracing of the boundary of the characters has reached the orlginal ~tarting point. If the coordinates are different, the next point (block 70) i8 searched.
If the vector direction i8 found (block 74) not ~o be in the 1 or 5 direction, the vector direction is tested for dlrectlons 2, 3 or 4 (block 82 of Fig. 9B). If the vector i8 found to correspond to one of these directions indicat~ng that the boundary ~ the character i8 extending in a downward direc-tion, the system will search in a direction 90 to the vector direction of the point sensed and in the direction 5 (F~g. 2) thereby mea~uring (block 84-Fig. 9B) the thickness of the character at this point. In meesuring the thickness, the system will count the number of binary one's sensed before finding a binary one to zero transition indicating the end of the portion of the character being sensed. As disclosed previously, located in the processor unit 32 (Fig. 6) are a plurality of counters 10855~4 represented by counters 48 and 49, each representing a predeter-mined character thickness. Thus, there may be a counter for each thickness of between 2 and 15 where the thickness i~ in terms of unit length. Each counter will count the number of times the thickness i8 found during the tracing of the character. The counters 48-49 corresponding to the thickness determined by the system will then be incremented by one (block 86) and the point will be tested (block 76) to detersnine if the coordinates of the point set a new limit.
If the result of block 82 is negative, the direction of the vectors is tested in direction~ 6, 7 or 8 (block 88) with the thickness being measured at r~ght angles to the vector direction and in direction 1 (block 90 of Fig. 9B) with the appropriate thickness counters48-49 being incremented (block 86) by one. This tracing of the boundary of the character 20 and the thickness measurement will continue until the last point in the boundary that i8 sensed has its coordinates corresponding to the first point ~easured (block 80) indicating that the traclng operation i8 complete. The thickness of the character is then 20 selected (block 92 of Fig. 96) by taking the highest count found in the thickness counters 48-49, inclusive, and storing the re8ult in the storage unit 41 (Fig. 6) A second boundary tracing of the lmage (Fig. 3) in storage unit 36 (Fig. 6) i8 then initiated (block 96) to reduce the boundary of the character 20 by replacing the character image shawn in Fig. 4 with the character image shwm in Fig. 5, which is the "left-hand" boundary of the character 20 (Flg. 1).
During thi~ second tracing of the character image, the thick-ness determined in block 92 and now stored in ~torage unit 41 (Fig. 6) w~ll be used to place each point that i~ found when moving in directions 2, 3 or 4 at a point in the storage unit 37 which will correspond to the left-hand boundary of the character 20. This point will be moved in direction 5 to the new location in the storage unit 37 which is equal to the thickness of the character. During this second tracing of the outer boundary of the image, the vector direction found i8 tested to detenmine in which direction the slope of the edge of the character 20 i8 proceeding ~o as to apply the above boundary conditions when the vectors indicate the slope to be in directions 2, 3 or 4.
The tracing operation i8 initiated (block 96) in the same manner as was previously described with respect to the first tracing operation with the operation of blocks 98, 100 and 102 of Fig.
9C perfonming the same function as blocks 70, 71 and 72, respectfully of Fig. 9A. The vector d~rection i8 tested (block 104) for directions 6, 7, or 8 to detenmine if the left-hand boundary of the thinned character image (Fig. 5) is being traced.
If it is found true, the coordinates of the point found are used to check (block 106) the same coordinates in the storage unit 37 to detenmine if a binary one has already been inserted in that location. If a binary one is found in that location, the system will start a new point search (block 98 of Fig. 9C) after going through block 128 of Fig. 9E testing for the start-ing point. This procedure insures that the thinnin8 operatlon does not create a new point location in the storage unit 37 that was not in the original image in storage unit 36.
If block 106 finds no point at that location in the picture in storage unit 37, the boundary point cuunter S0 (Fig.
which count~ the number of points in the boundary of the character i~ incremented by one (block 108 of Fig. 9D) and a binary one is then inserted (block 110) in the storage unit 37 at the same coordinates. The next point (block 98 of Fig. 9C) 0 i8 then found after going through block 128 of Fig. 9E. The point counter 50 is used in a succeeding tracing operation to insure that the boundary of the character i8 completely traced a~ will be explained more fully hereinafter.
If block 104 of Fig. 9C finds that the vector direc-tlon is not in directions 6, 7 or 8, the system will then move the coordinates of the point in the 5 direction (block 112) a distance equal to the thickness to determine ~f there i8 a one in that location in the image in storage unit 36. Thi8 procedure will check each point of the image for an edge point.
If a ~ne i~ not located, as a result of the operation of block 112, the thickness i~ reduced by one unit (block 114~of Fig. 9D) ~ntil the thickness ~s ~ero (block 116) indicating the point i8 an edge point. If a binary one i8 found when the thickness is not zero (block 116) indicating the point is not a part of the character, the system will search for the next point (block 90 of Fig. 9C). The direction to the next point i8 tested for continuation in the one direction (block 118 of Fig. 9D) or continuation in the five direction (block 120), both operations testing for the horizontal edge in the upper portion (block 118) and the lower portion (block 122) of the image. By identifying the horizontal edges of the character, the binary data requ1red to describe the character can be reduced by indicating the number of times the vector digit is repeated. This procedure also determines that portion of the outer boundary of the character which will be reduced by the thlckness of the character to establish the left-hand boundary of the character as will now be described.
If blocks 118 and 120 indicate that the direction to the next point under consideration is neither in direction five or one (Fig. 2), the point must be located in a vector directlon of either 2, 3, or 4 estsblishing that the right-hand boundary of the character 20 (Fig, 4) i8 being traced requiring that the point be positioned in the left-hand boundary of the picture in the storflge unlt 37. Block 122 tests the left-hand boundary po~ition of the point in the storage unit 37 by moving the coordinates of the point in the five direction equal to the value of the thickness stored in storage unit 41 (Fig. 6) to see if a one is already in that position. If it is, block 128 is tested and lf it is not the starting point, the next point 18 traced (block 98). If a one is not found in the tested position, the point counter 50 is incremented by one (block 124), the coordin-ates of the left-hand boundary points are ~tored in the width storage unit 40 (block 125) as the new width limit~, and a binary one i5 placed in the storage unit 37 (block 126-Fig 9E) corresponding to the left-hand boundary of the image in storage unit 36. This procedure i8 repeated until the coordinates of the last point traced coincide with the coordinstes of the starting point (block 128) indicating that the boundary (block 130) of theimage in the storage unit 36 has been traced.
The picture of the thinned character i8 then transferred to the ~torage unit 36 (block 132) replacing the origlnRl image. The bit counter 50 at this time will contain a count equal to the i number of points in the boundary of the thinned character which will be used to check the tracing of the boundary in a manner that will now be described.
The thinned character (Fig. 5) now stored in storage unlt 36 i8 te~ted to determine the number of bits required to describe the character when using the vector string as the descriptive method. The left-hand boundary tracing procedure of the thinned character i8 initiated (block 134 of Fig. 9E) and the first point i8 checked to determine if its coordinate~
are that of the starting point (block 136). The bit counter 50 i8 also checked to determine if its output i8 zero which would indicate th~t a boundary tracing of the thinned character has already taken place. This eliminates a problem found in tracing a character such as a "~' where the starting point may be reached before completing the tracing of the character. If the coordinate~ do not correspond to the ~tarting point or if 10855~4 counter ~0 is not zero (block 137) the next point is found ~-(block 138 of Fig. 9F). Block 140 detenmine6 the vector direction to the next point, the system moving to thst point (block 142). The binary digit representing the vec~or ~8 then ~-stored in storage unit 38, the vector counter 51 (Fig. 6) i8 incremented by one (block 146), the point counter 50 i8 decremented by one (block 147), and the vector counter 51 is checked (block 148) to determine if the count at this t~me i8 less than a predetermined number n, which in the present example i8 256, n being the capAcity of the storage unit 38.
If the count in the counter 51 is less than n, the sequence i8 rrepeated until the thinned character has been traced and the , storage unit 38 will contain binary data representing the ', vectors in the left-hand boundary of the thlnned character.
The vector number counter 51 wlll contain a count equal to , the number of vectors ln the thinned character (Fig. S) at thl~
time.
Block 150 will inltiate a trace of the vector string , stored in storage unit 38 to detenmine the number of corners '~ 20 in the thinned chsracter. As previously described, each change of vector direction between two ~uccessive vect3rs in the vector string is a corner and a character can be described by knowing the cartesian coordinates of each corner listed in the sequence that they are found. Block 152 of Fig. 9G will compare each 3 vector with respect to the last to determine if a corner exists.
; A corner,exlsts when there i5 no agreement between two successive "
' ~ O 8 ~ ~ 4 vectors~ If there i8 agreement, the vectors will continue (block 150) to be compared until a corner i~ found. Once found, the location of the corner in tenms of its cartesian coordinates (block 154), is stored (block 156) in ~torage unit 29 (Fig. 6) and the corner coordinate~ are cumpared with the first corner (block 158) to find the starting point. If the coordinates are not the starting point, the next two vectors are compared (block 150) to find the next corner. This process continues until the starting point is reached (block 158).
At this time the number of binary bits representing the corners stored in storage unit 29 is determined (block 160) utilizing the arithmetic unit 46 (Fig 6) in a manner well known in the art with the result stored in the storage unit 44.
Additionally, the count number in the counter 51 is multiplied by three in the arithmet~c unit 46 to produce the number of bit8 required to define the characters ln terms offthe vector string. This result i8 stored in storage unit 43. The number ; of blnary bits in the picture of the thinned character is determined ~block 164) by mult~plying the height ~imes the width using the limits stored in storage units 39, 40 durlng the tracing of the charac~er ~n block 78.and in block 125.
Thi~ result is stored in storage unit 42 (Fig. 6). The outputs of the data stored in storage units 29, 42 and 43 are then outputted to a minimum binary comparator 47 (Fig. 6) to compare and select (blocks 166, 168, and 170 of Fig. 9H) which of the storage unitscontains the m~nimNm number of binary bits. The comparator 47 will compare two of the values in storage units 29, 42 and 43 determining the minimum of the two and then compare this result with the third value. The binary data selected i8 then outputted (blocks 172, 174 and 176) o~er line 177 to a recording unit 34 (Fig. 1) for recording the selected character representing data in a recording medium for further processing. The processing-,of the document 22 (Fig. 1) in accordance with the present invention i~ then continued until completed (block 178).
While the invention has been shown and described in tenms of a preferred embodiment thereof, it will be understood that this invention i8 not limited to this particular embod~ment and that many changes and modifications may be made without departing fro~ the true spirit and scope of the invention as defined in the appended claims. It is therefore obvious that the right-hand boundary of the character may be developed for u~e in determining the minimum amount of data which describes the character.
Claims (20)
1. In a pattern recognition process including the steps of scanning a pattern and digitizing elemental areas of the pattern into pattern digital bits and background digital bits, a pattern description method comprising the steps of (a) storing in a first storage means the pattern digital bits representing a character thinned through a boundary reduction technique;
(b) forming vector representation of line segments connecting adjacent pattern digital bits stored in said first storage means;
(c) determining the amount of digital bits in said vector representation;
(d) determining the amount of digital bits in said pattern digital bits and background digital bits;
(e) and selecting the minimum amount of digital bits determined in steps (c) and (d) for recording on a recording medium.
(b) forming vector representation of line segments connecting adjacent pattern digital bits stored in said first storage means;
(c) determining the amount of digital bits in said vector representation;
(d) determining the amount of digital bits in said pattern digital bits and background digital bits;
(e) and selecting the minimum amount of digital bits determined in steps (c) and (d) for recording on a recording medium.
2. The method of claim 1 wherein said step of storing the pattern bits representing the thinned character comprises the steps of (a) determining the thickness of the pattern;
(b) scanning the pattern digital bits comprising the outer boundary of the pattern;
(c) storing a portion of the outer boundary of the pattern in the first storage area;
(d) and replacing the remaining portion of the outer boundary 2 (concluded) in the first storage area with the inner boundary of the pattern by reducing the remaining portion of the outer boundary by the thickness of the pattern.
(b) scanning the pattern digital bits comprising the outer boundary of the pattern;
(c) storing a portion of the outer boundary of the pattern in the first storage area;
(d) and replacing the remaining portion of the outer boundary 2 (concluded) in the first storage area with the inner boundary of the pattern by reducing the remaining portion of the outer boundary by the thickness of the pattern.
3. The method of claim 2 wherein said step of determining the thickness of the pattern comprises the steps of (a) scanning the pattern digital bits forming the outer boundary of the pattern;
(b) sensing the thickness of the pattern at each digital bit in the outer boundary of the pattern;
(c) counting the number of different thicknesses sensed;
(d) and selecting the thickness having the highest number of counts as the thickness of the pattern.
(b) sensing the thickness of the pattern at each digital bit in the outer boundary of the pattern;
(c) counting the number of different thicknesses sensed;
(d) and selecting the thickness having the highest number of counts as the thickness of the pattern.
4. The method of claim 1 wherein the step of determining the amount of digital bits in the vector representation com-prises the steps of (a) assigning a digital code to the vector representation representing the direction of said line segments connecting succeeding digital bits in the reduced boundary of the pattern;
(b) counting the number of vectors in the reduced boundary of the pattern;
(c) and multiplying the number of vectors by the number of digital bits in the digital code.
(b) counting the number of vectors in the reduced boundary of the pattern;
(c) and multiplying the number of vectors by the number of digital bits in the digital code.
5. The method of claim 2 wherein the step of determining the amount of digital bits in the pattern in said first storage area comprises the steps of (a) assigning cartesian coordinates to the location of each digital bit in said first storage area;
(b) storing the maximum coordinates in the x and y direction of the digital bits in the reduced boundary of the pattern;
(c) determining the number of digital bits in the height and width of the pattern from the stored coordinates;
(d) and multiplying the digital bits in the height and the width of the pattern to produce the number of bits in the pattern.
(b) storing the maximum coordinates in the x and y direction of the digital bits in the reduced boundary of the pattern;
(c) determining the number of digital bits in the height and width of the pattern from the stored coordinates;
(d) and multiplying the digital bits in the height and the width of the pattern to produce the number of bits in the pattern.
6. The method of claim 5 further comprising the steps of (a) determining the location of the corners in the reduced boundary of the pattern;
(b) assigning cartesian coordinates to the location of each corner in the reduced boundary of the pattern;
(c) storing the digital bits representing the cartesian coordinates of each of the corners in a second storage area;
(d) determining the number of digital bits in said second storage area;
(e) and selecting the minimum amount of digital bits determined with respect to the vector representation, the pattern and the corners.
(b) assigning cartesian coordinates to the location of each corner in the reduced boundary of the pattern;
(c) storing the digital bits representing the cartesian coordinates of each of the corners in a second storage area;
(d) determining the number of digital bits in said second storage area;
(e) and selecting the minimum amount of digital bits determined with respect to the vector representation, the pattern and the corners.
7. The method of claim 6 wherein the step of selecting the minimum amount of digital bits in the vector representation, the pattern and the corners comprises the steps of (a) storing the number of digital bits in the vector representation in a third storage area;
(b) storing the number of digital bits in the pattern in a fourth storage area;
(c) storing the number of digital bits in the corners of the reduced boundary of the pattern in a fifth storage area;
(d) and comparing the number of digital bits stored in steps (a), (b) and (c) to determine the minimum number of digital bits in said storage areas.
(b) storing the number of digital bits in the pattern in a fourth storage area;
(c) storing the number of digital bits in the corners of the reduced boundary of the pattern in a fifth storage area;
(d) and comparing the number of digital bits stored in steps (a), (b) and (c) to determine the minimum number of digital bits in said storage areas.
8. A method for determining the minimum data required to describe a character comprising the steps of (a) scanning the character to determine an array of digital signals representative of the configuration of the character;
(b) storing the digital signals in a first storage unit;
(c) processing the digital signal in the outer boundary of the configuration of the character to determine the thickness of the character;
(d) examining the digital signals in the outer boundary of the configuration to reduce the configuration to the inner boundary of the configuration using the thickness of the character to determine the inner boundary of a portion of 8 (concluded) the outer boundary of the character;
(e) generating a plurality of coded vector representations of the direction of the line segments interconnecting the digital signals in the inner boundary of the configuration;
(f) determining the number of digital signals in the coded vector representation;
(g) determining the number of digital signals in the inner boundary configuration of the character;
(h) and selecting the least number of digital signals deter-mined in steps (f) ant (g).
(b) storing the digital signals in a first storage unit;
(c) processing the digital signal in the outer boundary of the configuration of the character to determine the thickness of the character;
(d) examining the digital signals in the outer boundary of the configuration to reduce the configuration to the inner boundary of the configuration using the thickness of the character to determine the inner boundary of a portion of 8 (concluded) the outer boundary of the character;
(e) generating a plurality of coded vector representations of the direction of the line segments interconnecting the digital signals in the inner boundary of the configuration;
(f) determining the number of digital signals in the coded vector representation;
(g) determining the number of digital signals in the inner boundary configuration of the character;
(h) and selecting the least number of digital signals deter-mined in steps (f) ant (g).
9. The method of claim 8 wherein the step of examining the digital signals to reduce the configuration to the inner boundary comprises the steps of (a) scanning the digital signals in the outer boundary of the configuration in said first storage unit;
(b) transferring the digital signals sensed in a portion of the outer boundary of the configuration to a second storage unit;
(c) and storing in the second storage unit the digital signals in the reduced portion of the remainder of the configuration by reducing the location of the digital signals in the remaining portion of the outer boundary by the thickness of the character.
(b) transferring the digital signals sensed in a portion of the outer boundary of the configuration to a second storage unit;
(c) and storing in the second storage unit the digital signals in the reduced portion of the remainder of the configuration by reducing the location of the digital signals in the remaining portion of the outer boundary by the thickness of the character.
10. The method of claim 8 wherein said step of processing the digital signals to determine the thickness of the character 10 (concluded) comprises the steps of (a) scanning the digital signals in the outer boundary of the character;
(b) sensing the number of digital signals in the thickness of the character at each digital signal in the outer boundary of the character;
(c) counting the number of different thicknesses sensed in step (b);
(d) and selecting the thickness having the highest number count as the thickness of the character.
(b) sensing the number of digital signals in the thickness of the character at each digital signal in the outer boundary of the character;
(c) counting the number of different thicknesses sensed in step (b);
(d) and selecting the thickness having the highest number count as the thickness of the character.
11. The method of claim 9 wherein the step of determining the amount of digital signals in the coded vector representation comprises the steps of (a) determining the number of digital signals in the code assigned to the vector representation;
(b) scanning the digital signals in the inner boundary of the character to develop the coded vector representation, each representing the direction of the line segments interconnecting adjacent digital signals in the boundary;
(c) counting the number of coded vectors in the inner boundary of the character;
(d) and multiplying the number of vectors by the number of digital signals in the code.
(b) scanning the digital signals in the inner boundary of the character to develop the coded vector representation, each representing the direction of the line segments interconnecting adjacent digital signals in the boundary;
(c) counting the number of coded vectors in the inner boundary of the character;
(d) and multiplying the number of vectors by the number of digital signals in the code.
12. The method of claim 11 wherein the step of determining the number of digital signals in the inner boundary configuration of the character comprises the steps of (a) assigning cartesian coordinates to the location of each digital signal in the second storage unit;
(b) scanning the inner boundary configuration of the character;
(c) storing the digital value of the cartesian coordinates of the highest and widest digital signals in the inner boundary of configuration of the character;
(d) determining the number of digital signals in the height and widith of the inner boundary configuration of the character from the digital value of the stored coordinates;
(e) and multiplying the number of digital signals in the height ant width to produce the number of digital signals in the configuration of the character.
(b) scanning the inner boundary configuration of the character;
(c) storing the digital value of the cartesian coordinates of the highest and widest digital signals in the inner boundary of configuration of the character;
(d) determining the number of digital signals in the height and widith of the inner boundary configuration of the character from the digital value of the stored coordinates;
(e) and multiplying the number of digital signals in the height ant width to produce the number of digital signals in the configuration of the character.
13. The method of claim 12 further comprising the steps of (a) comparing adjacent coded vector representation to determine the location of the corners in the inner boundary of the character;
(b) assigning cartesian coordinates to the location in said second storage areas of the corners in the inner boundary of he character;
(c) storing the digital value of the cartesian coordinates of the corners in a third storage unit;
(d) determining the number of digital signals in the digital values stored in said third storage area;
13 (concluded) (e) and selecting the minimum number of the digital signals to determine with respect to the vector representation, the character and the corners.
(b) assigning cartesian coordinates to the location in said second storage areas of the corners in the inner boundary of he character;
(c) storing the digital value of the cartesian coordinates of the corners in a third storage unit;
(d) determining the number of digital signals in the digital values stored in said third storage area;
13 (concluded) (e) and selecting the minimum number of the digital signals to determine with respect to the vector representation, the character and the corners.
14. The method of claim 13 wherein the step of selecting the minimum number of digital signals comprises the steps of (a) storing the number of digital signals in the vector representation in a fourth storage unit;
(b) storing the number of digital signals in the configuration of the character in a fifth storage unit;
(c) storing the number of digital signal in the corners in a sixth storage unit;
(d) and comparing the number of binary bits stored in said fourth, fifth and sixth storing areas to determine the minimum number of binary bits in said storage areas.
(b) storing the number of digital signals in the configuration of the character in a fifth storage unit;
(c) storing the number of digital signal in the corners in a sixth storage unit;
(d) and comparing the number of binary bits stored in said fourth, fifth and sixth storing areas to determine the minimum number of binary bits in said storage areas.
15. A method of describing characters in a document comprising the steps of (a) scanning the characters to derive an array of binary bits representative of the configuration of the character whereby a character binary bit is distinguished from a background binary bit;
(b) storing the binary bits in a first matrix of storage areas in a processor;
(c) sensing the binary bits of a predetermined group of said storage areas starting from a storage area containing a 15 (continued) first character binary bit located in the outer boundary of the character and sensing the binary bits in the remaining areas within the group to locate the next character binary bit in the boundary of the character;
( d) sensing the number of character binary bits in the character at right angles to the character binary bit sensed in the outer boundary of the character to determine the thickness of the character;
(e) counting each time a number of character bits in the thickness of the character is sensed;
(f) repeating steps (c), (d) and (e) until all of the character binary bits in the outer boundary have been sensed and the thickness of the character sensed have been counted;
(g) selecting the highest number thicknesses counted as the thickness of the character;
(h) scanning the outer boundary of the character in said first storage area;
(i) storing the portions of the outer boundary scanned corres-ponding to the left-hand boundary of the character in a second storage area in a processor;
(j) reducing the remaining portion of the outer boundary scanned corresponding to the right-hand boundary of the character by the thickness of the character to produce the left-hand boundary of the character;
15 (concluded) (k) storing the left-hand boundary configuration of the character in said second storage area;
(l) sensing the character binary bits in the left-hand boundary of the character stored in said second storage area;
(m) generating a coded vector representation of the direction between the character binary bits in the left-hand boundary sensed;
(n) determining the number of binary bits in the coded vector representation;
(o) determining the number of binary bits in the left-hand boundary configuration of the character;
(p) and selecting the least number of binary bits determined in steps (n) and (o) for recording on a recording medium.
(b) storing the binary bits in a first matrix of storage areas in a processor;
(c) sensing the binary bits of a predetermined group of said storage areas starting from a storage area containing a 15 (continued) first character binary bit located in the outer boundary of the character and sensing the binary bits in the remaining areas within the group to locate the next character binary bit in the boundary of the character;
( d) sensing the number of character binary bits in the character at right angles to the character binary bit sensed in the outer boundary of the character to determine the thickness of the character;
(e) counting each time a number of character bits in the thickness of the character is sensed;
(f) repeating steps (c), (d) and (e) until all of the character binary bits in the outer boundary have been sensed and the thickness of the character sensed have been counted;
(g) selecting the highest number thicknesses counted as the thickness of the character;
(h) scanning the outer boundary of the character in said first storage area;
(i) storing the portions of the outer boundary scanned corres-ponding to the left-hand boundary of the character in a second storage area in a processor;
(j) reducing the remaining portion of the outer boundary scanned corresponding to the right-hand boundary of the character by the thickness of the character to produce the left-hand boundary of the character;
15 (concluded) (k) storing the left-hand boundary configuration of the character in said second storage area;
(l) sensing the character binary bits in the left-hand boundary of the character stored in said second storage area;
(m) generating a coded vector representation of the direction between the character binary bits in the left-hand boundary sensed;
(n) determining the number of binary bits in the coded vector representation;
(o) determining the number of binary bits in the left-hand boundary configuration of the character;
(p) and selecting the least number of binary bits determined in steps (n) and (o) for recording on a recording medium.
16. The method of claim 15 wherein the step of determining the number of binary bits in the coded vector representation comprises the steps of (a) storing in a third storage area of the processor a predetermined numbered order of angular directions between each succeeding character binary bits sensed in said predetermined group of storage areas;
(b) repeating step (a) until all the angular directions between each succeeding character binary bit in the left-hand boundary of the character are stored in the third storage area;
16 (concluded) (c) counting the number of angular directions in the third storage area;
(d) and multiplying the number of angular directions by the number of binary bits in the numbered order of angular directions.
(b) repeating step (a) until all the angular directions between each succeeding character binary bit in the left-hand boundary of the character are stored in the third storage area;
16 (concluded) (c) counting the number of angular directions in the third storage area;
(d) and multiplying the number of angular directions by the number of binary bits in the numbered order of angular directions.
17. The method of claim 16 wherein the step of determining the number of binary bits in the left-hand configuration of the character comprises the steps of (a) assigning cartesian coordinated to the location of each character binary bits in the second storage area;
(b) scanning the character bits in the left-hand boundary of the character in the second storage area;
(c) storing the digital value of the cartesian coordinates of the highest and widest character binary bit scanned in the left hand configuration of the character in a fourth ant a fifth storage area of the processor, respectively;
(d) determining the number of binary bits in the height and width of the character from the digital value stored in the fourth and fifth storage areas;
(e) and multiplying the number of binary bits in the height and width of the character to derive the number of binary bits in the character.
(b) scanning the character bits in the left-hand boundary of the character in the second storage area;
(c) storing the digital value of the cartesian coordinates of the highest and widest character binary bit scanned in the left hand configuration of the character in a fourth ant a fifth storage area of the processor, respectively;
(d) determining the number of binary bits in the height and width of the character from the digital value stored in the fourth and fifth storage areas;
(e) and multiplying the number of binary bits in the height and width of the character to derive the number of binary bits in the character.
18. The method of claim 17 further comprising the steps of (a) comparing adjacent numbered order of angular directions stored in the third storage area to determine a change of direction in the left-hand boundary of the character;
(b) assigning cartesian coordinates to the location in the second storage area of each change in direction;
(c) storing the digital value of the cartesian coordinates of each direction change in a seventh storage area in the processor;
(d) determining the number of binary bits in the digital values stored in the seventh storage area;
(e) and selecting the minimum number of binary bits determined with respect to the angular directions, the character and the direction changes.
(b) assigning cartesian coordinates to the location in the second storage area of each change in direction;
(c) storing the digital value of the cartesian coordinates of each direction change in a seventh storage area in the processor;
(d) determining the number of binary bits in the digital values stored in the seventh storage area;
(e) and selecting the minimum number of binary bits determined with respect to the angular directions, the character and the direction changes.
19. The method of claim 18 wherein the step of selecting the minimum number of binary bits comprises the steps of (a) storing the number of binary bits in the angular directions in a eighth storage area of the processor;
(b) storing the number of binary bits in the character in a ninth storage area of the processor;
(c) storing the number of binary bits in the angular direction changes in a tenth storage area of the processor;
(d) and comparing the number of binary bits stored in said eighth, ninth and tenth storage areas to determine the minimum number of binary bits in said storage areas.
(b) storing the number of binary bits in the character in a ninth storage area of the processor;
(c) storing the number of binary bits in the angular direction changes in a tenth storage area of the processor;
(d) and comparing the number of binary bits stored in said eighth, ninth and tenth storage areas to determine the minimum number of binary bits in said storage areas.
20, In a pattern recognition system having means for digitizing an analog pattern into pattern digital bits and background digital bits, means for determining the minimum amount of data describing the pattern comprising (a) means for storing pattern digital bits representing a character thinned through a boundary reduction technique;
(b) means for forming vector representation of line segments connecting adjacent pattern digital bits in the boundary of the character stored in said storage means;
(c) first means for generating the amount of digital bits in said vector representation;
(d) second means for generating the amount of digital bits in said pattern and background digital bits;
(e) and means for selecting the minimum amount of digital bits between the vector representation and the pattern and background digital bits.
(b) means for forming vector representation of line segments connecting adjacent pattern digital bits in the boundary of the character stored in said storage means;
(c) first means for generating the amount of digital bits in said vector representation;
(d) second means for generating the amount of digital bits in said pattern and background digital bits;
(e) and means for selecting the minimum amount of digital bits between the vector representation and the pattern and background digital bits.
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| US05/759,604 US4087788A (en) | 1977-01-14 | 1977-01-14 | Data compression system |
| US759,604 | 1977-01-14 |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CA1085514A true CA1085514A (en) | 1980-09-09 |
Family
ID=25056283
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CA287,295A Expired CA1085514A (en) | 1977-01-14 | 1977-09-22 | Data compression system |
Country Status (6)
| Country | Link |
|---|---|
| US (1) | US4087788A (en) |
| JP (1) | JPS5388533A (en) |
| CA (1) | CA1085514A (en) |
| DE (1) | DE2801536C2 (en) |
| FR (1) | FR2377670A1 (en) |
| GB (1) | GB1567287A (en) |
Families Citing this family (52)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US4208651A (en) * | 1978-05-30 | 1980-06-17 | Sperry Corporation | Fingerprint identification by ridge angle and minutiae recognition |
| US4580231A (en) * | 1978-09-15 | 1986-04-01 | Alphatype Corporation | Ultrahigh resolution photocomposition system employing electronic character generation from magnetically stored data |
| US4264808A (en) * | 1978-10-06 | 1981-04-28 | Ncr Corporation | Method and apparatus for electronic image processing of documents for accounting purposes |
| US6393151B1 (en) | 1978-10-13 | 2002-05-21 | Agency Of Industrial Science And Technology | Pattern reading system |
| US4307377A (en) * | 1979-11-09 | 1981-12-22 | Bell Telephone Laboratories, Incorporated | Vector coding of computer graphics material |
| JPS5846460A (en) * | 1981-09-11 | 1983-03-17 | Matsushita Graphic Commun Syst Inc | Formation system for contour picture |
| US4520506A (en) * | 1981-10-20 | 1985-05-28 | Harris Corporation | Method and system for compression and reconstruction of cultural data for use in a digital moving map display |
| US4538299A (en) * | 1981-12-04 | 1985-08-27 | International Remote Imaging Systems, Inc. | Method and apparatus for locating the boundary of an object |
| US4490848A (en) * | 1982-03-31 | 1984-12-25 | General Electric Company | Method and apparatus for sorting corner points in a visual image processing system |
| US4493105A (en) * | 1982-03-31 | 1985-01-08 | General Electric Company | Method and apparatus for visual image processing |
| EP0131676B1 (en) * | 1983-07-04 | 1989-08-02 | URW Software & Type GmbH | Method for automatically digitizing the contours of line graphics, e.g. characters |
| US4598411A (en) * | 1984-07-17 | 1986-07-01 | Allied Corporation | On-the-fly data compression system |
| JPS61188681A (en) * | 1985-02-15 | 1986-08-22 | Matsushita Electric Ind Co Ltd | Character recognition device |
| US4777620A (en) * | 1985-02-20 | 1988-10-11 | Elscint Ltd. | Data compression system |
| JPS61246877A (en) * | 1985-04-25 | 1986-11-04 | Canon Inc | Method for converting linear graphic |
| JPS6282484A (en) * | 1985-10-07 | 1987-04-15 | Canon Inc | Method for converting linear graphic form |
| US5365599A (en) * | 1985-10-07 | 1994-11-15 | Canon Kabushiki Kaisha | Method and system of converting delineative pattern |
| DE3716787A1 (en) * | 1986-05-19 | 1987-11-26 | Ricoh Kk | CHARACTER RECOGNITION METHOD |
| US5097517A (en) * | 1987-03-17 | 1992-03-17 | Holt Arthur W | Method and apparatus for processing bank checks, drafts and like financial documents |
| US4837842A (en) * | 1986-09-19 | 1989-06-06 | Holt Arthur W | Character and pattern recognition machine and method |
| US4817187A (en) * | 1987-02-19 | 1989-03-28 | Gtx Corporation | Apparatus and method for vectorization of incoming scanned image data |
| US5020124A (en) * | 1989-10-10 | 1991-05-28 | Unisys Corporation | Method and apparatus for detecting document size in an imaging system |
| US5287416A (en) * | 1989-10-10 | 1994-02-15 | Unisys Corporation | Parallel pipelined image processor |
| US5140444A (en) * | 1989-10-10 | 1992-08-18 | Unisys Corporation | Image data processor |
| US5095374A (en) * | 1989-10-10 | 1992-03-10 | Unisys Corporation | Method and apparatus for lossless compression and decompression of image data |
| US5093871A (en) * | 1989-10-10 | 1992-03-03 | Unisys Corporation | Method and apparatus for effecting background suppression of image data |
| US5048104A (en) * | 1989-10-10 | 1991-09-10 | Unisys Corporation | Method and apparatus for transposing image data |
| US5029226A (en) * | 1989-10-10 | 1991-07-02 | Unisys Corporation | Method and apparatus for effecting spot/void filtering of image data |
| US5305398A (en) * | 1989-10-10 | 1994-04-19 | Unisys Corporation | Method and apparatus for scaling image data |
| US5055919A (en) * | 1989-10-10 | 1991-10-08 | Unisys Corporation | Apparatus for image data transposition and compression/decompression |
| US5091975A (en) * | 1990-01-04 | 1992-02-25 | Teknekron Communications Systems, Inc. | Method and an apparatus for electronically compressing a transaction with a human signature |
| US5049881A (en) * | 1990-06-18 | 1991-09-17 | Intersecting Concepts, Inc. | Apparatus and method for very high data rate-compression incorporating lossless data compression and expansion utilizing a hashing technique |
| US5157740A (en) * | 1991-02-07 | 1992-10-20 | Unisys Corporation | Method for background suppression in an image data processing system |
| US5227590A (en) * | 1991-05-17 | 1993-07-13 | Ncr Corporation | Handwriting capture device |
| US5267332A (en) * | 1991-06-19 | 1993-11-30 | Technibuild Inc. | Image recognition system |
| CA2077271C (en) * | 1991-12-13 | 1998-07-28 | David J. Craft | Method and apparatus for compressing data |
| US5396228A (en) * | 1992-01-16 | 1995-03-07 | Mobile Telecommunications Technologies | Methods and apparatus for compressing and decompressing paging data |
| US5371499A (en) * | 1992-02-28 | 1994-12-06 | Intersecting Concepts, Inc. | Data compression using hashing |
| US5406278A (en) * | 1992-02-28 | 1995-04-11 | Intersecting Concepts, Inc. | Method and apparatus for data compression having an improved matching algorithm which utilizes a parallel hashing technique |
| US5353024A (en) * | 1992-05-01 | 1994-10-04 | Intersecting Concepts, Inc. | Method for data compression having an improved encoding algorithm which utilizes a token stacking technique |
| US5448044A (en) * | 1993-04-30 | 1995-09-05 | Microbilt Corporation | Signature capture pad for point of sale system |
| US5563595A (en) * | 1993-12-23 | 1996-10-08 | International Business Machines Corporation | Method and apparatus for compressing data |
| DE69512824T2 (en) * | 1994-04-22 | 2000-01-27 | Victor Company Of Japan | Compression and decompression processes for multi-dimensional multi-color images |
| JP3415270B2 (en) * | 1994-06-03 | 2003-06-09 | ソニー株式会社 | Image signal encoding method and decoding method |
| US5818368A (en) * | 1997-04-18 | 1998-10-06 | Premier Research, Llc | Method and apparatus for lossless digital data compression |
| US5956428A (en) * | 1997-05-19 | 1999-09-21 | Ausbeck, Jr.; Paul James | Stroke code for image partitions |
| US6728412B1 (en) | 1999-10-29 | 2004-04-27 | S.V.V. Technology Innovations, Inc. | Method and apparatus for on-the-fly image coding |
| US6714145B1 (en) | 2002-09-26 | 2004-03-30 | Richard Marques | Method and apparatus for integer-based encoding and decoding of bits |
| US7787708B2 (en) * | 2005-06-07 | 2010-08-31 | Ids Scheer Aktiengesellschaft | Systems and methods for rendering text within symbols |
| KR101174676B1 (en) * | 2010-11-19 | 2012-08-17 | 주식회사 고영테크놀러지 | Method and apparatus of profiling a surface |
| US8779950B2 (en) | 2012-03-05 | 2014-07-15 | Dcba, Llc | Command encoded data compression |
| US9543980B2 (en) | 2014-10-10 | 2017-01-10 | Massachusettes Institute Of Technology | Systems and methods for model-free compression and model-based decompression |
Family Cites Families (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US3394352A (en) * | 1965-07-22 | 1968-07-23 | Electronic Image Systems Corp | Method of and apparatus for code communication |
| GB1171627A (en) * | 1966-10-07 | 1969-11-26 | Post Office | Improvements in or relating to Character Recognition Machines |
| CA1005916A (en) * | 1972-08-16 | 1977-02-22 | International Business Machines Corporation | Video compaction for printed text |
| US3987412A (en) * | 1975-01-27 | 1976-10-19 | International Business Machines Corporation | Method and apparatus for image data compression utilizing boundary following of the exterior and interior borders of objects |
| DE2516332A1 (en) * | 1975-04-15 | 1976-10-28 | Siemens Ag | METHOD OF ENCODING ELECTRICAL SIGNALS OBTAINED WHEN SCANNING A GRAPHIC PATTERN WITH CONTENT MIXED FROM TEXT AND IMAGE |
| US4020463A (en) * | 1976-02-27 | 1977-04-26 | Recognition Equipment Incorporated | Apparatus and a method for storage and retrieval of image patterns |
-
1977
- 1977-01-14 US US05/759,604 patent/US4087788A/en not_active Expired - Lifetime
- 1977-09-22 CA CA287,295A patent/CA1085514A/en not_active Expired
-
1978
- 1978-01-11 GB GB1049/78A patent/GB1567287A/en not_active Expired
- 1978-01-12 JP JP158678A patent/JPS5388533A/en active Granted
- 1978-01-13 FR FR7800879A patent/FR2377670A1/en active Granted
- 1978-01-14 DE DE2801536A patent/DE2801536C2/en not_active Expired
Also Published As
| Publication number | Publication date |
|---|---|
| GB1567287A (en) | 1980-05-14 |
| FR2377670B1 (en) | 1984-09-21 |
| US4087788A (en) | 1978-05-02 |
| JPS5388533A (en) | 1978-08-04 |
| JPS6239462B2 (en) | 1987-08-24 |
| DE2801536C2 (en) | 1983-05-05 |
| DE2801536A1 (en) | 1978-07-20 |
| FR2377670A1 (en) | 1978-08-11 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CA1085514A (en) | Data compression system | |
| US4208651A (en) | Fingerprint identification by ridge angle and minutiae recognition | |
| US3999161A (en) | Method and device for the recognition of characters, preferably of figures | |
| US4075604A (en) | Method and apparatus for real time image recognition | |
| US4310827A (en) | Device for extracting a density as one of pattern features for each feature point of a streaked pattern | |
| US4712248A (en) | Method and apparatus for object identification | |
| US4047154A (en) | Operator interactive pattern processing system | |
| US4272756A (en) | Method of pattern recognition and apparatus thereof | |
| US4013999A (en) | Single read station acquisition for character recognition | |
| US4307377A (en) | Vector coding of computer graphics material | |
| US4288779A (en) | Method and apparatus for character reading | |
| EP0184775A2 (en) | Conversion of an image represented by a field of pixels in a gray scale to a field of pixels in binary scale | |
| JPH08185521A (en) | Mobile object counter | |
| EP0236781A2 (en) | Image data processing method and apparatus therefor | |
| EP0113086A2 (en) | Image data processing system | |
| US4700402A (en) | Input method for graphic pattern data | |
| US5386482A (en) | Address block location method and apparatus | |
| US4130819A (en) | Optical character recognition device | |
| JPH0664614B2 (en) | Hierarchical structured template matching method | |
| EP0435099A2 (en) | A method of selecting characteristics data for a data processing system | |
| US5825923A (en) | Method for performing character recognition on a pixel matrix | |
| US5313529A (en) | System for representing line drawings by polygonal lines | |
| JP3303748B2 (en) | Registration method of reference one-dimensional data string, image recognition method, registration device, image recognition device, and recording medium | |
| JP3026850B2 (en) | Automatic tracking device | |
| GB1525856A (en) | Image pattern selection and location method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| MKEX | Expiry |