WO1994003872A1

WO1994003872A1 - Inspection system using compressed image data

Info

Publication number: WO1994003872A1
Application number: PCT/US1993/007175
Authority: WO
Inventors: Jesse C. Booher; Thomas B. Sprinkle
Original assignee: Pressco Technology, Inc.
Priority date: 1992-07-31
Filing date: 1993-07-30
Publication date: 1994-02-17
Also published as: AU4995393A

Abstract

A video inspection system which accomplishes its bit-mapped graphical information is compressed by converting it to a format within the frequency domain (30). Compressed data is analyzed against data in a database to determine acceptability of the specimen (32). Specimens for which an acceptability determination is not completed by this comparison are subject to a more conventional analysis of the bit-mapped data corresponding to the earlier analyzed compressed data (32). This system is applied both to discrete specimen analysis as well as continuous specimen analysis such as occurs in web inspections.

Description

INSPECTION SYSTEM USING COMPRESSED IMAGE DATA

Background of the Invention

This application pertains to the art of automated inspection, and more particularly to product/process inspection, employing image data previously subjected to data compression.

The invention is particularly applicable to high speed video inspection and will be described with particular reference thereto, although it will be appreciated that the invention has broader application in analysis and cataloging of any data compressed into a frequency domain constituent, such as real-time analysis of a compressed signal data stream.

Automated video inspection is becoming increasingly popular for use in manufacturing. Automated inspection allows for objective analysis of specimen acceptability with an integrity and rate of speed not practicable by human inspection. Earlier inspection systems utilized video cameras to capture digitized specimen images. These captured images would be subjected to numeric-based analysis to determine specimen acceptability in accordance with preset standards.

Earlier acceptability testing was typically accomplished by two basic techniques. In the first, pattern matching an entire digitized bit-mapped image, or portion thereof, would be compared against preselected criteria to determine acceptability. In the second, algorithms would be performed on image edges or transition points, and this transition data would be utilized for comparison. Both earlier systems required substantial processing power to accommodate the analysis in real time. The processing powe- requirements increase with the increased resolution of the captured image. Further, the systems were extremely difficult to employ in continuous stream or web inspections, in which 100% surface inspection is desirable and a large variety of aberration patterns are possible.

An area traditionally distinct from video inspection is that of modulated signal transmission for direct digital information transfer. The data transmission industry is becoming increasingly aware of potential markets associated with real-time transmission of both still life and moving picture video, in addition to presently ubiquitous audio. However, typical transmission media employed for point-to-point communication, such as a telephone line, has insufficient bandwidth for transmission of real-time video signals in conventional bit-mapped video format.

The tension between limited bandwidth associated with conventional transmission media, and the desire for real-time video transmission, has led to innovation in data coding and algorithmic manipulation. Such algorithms and coding serve to reduce the amount of bandwidth necessary to communicate the data. Various data compression implementations exist which are inclusive of a baseline "lossy" approach and an extended "lossless" approach. A true lossless compression is somewhat rare. Lossless algorithms do not provide great compression ratios. A degree of image degradation associated with compression and decompression is a function of a degree of compression, image properties, and complexity of the compression algorithm.

A present standard for still image compression is referred to as JPEG ("Joint Photographic Experts Group") . A system currently in development for motion picture data compression is referred to as MPEG ("Moving Picture Experts Group") for full-motion compression on digital storage medium ("DSM") . An adaptation of JPEG to moving pictures is also currently in commercial use. An additional standard for video conferencing is provided by the Committee on International Telephony & Telegraphy ("CCITT") recommendation H.261.

By way of example, the JPEG based algorithm is comprised of transform-based image coding. A color image may be represented in different color systems. Those in wide use today include RGB (red, green, and blue) which is typically employed in the computer industry; YUV (Y for luminance or brightness U and V for color difference signals YR and YB, respectively) employed in the television industry; and CMYK ("Cyan, Magenta, Yellow and Black") employed in the printing industry. Within each color system, the constituent parts are referred to as components. JPEG encoding acts upon redundancies and trends found in the bit-mapped image of a representative color scheme. JPEG encoding is accomplished today by either software manipulation, hardware manipulation, or a hybrid combination thereof.

A more detailed background of JPEG, MPEG, and CCITT H.261 data compression may be found in Video Compression Makes Big Gains, IEEE Spectrum, October, 1991, p. 16-19. Summarizing briefly, each component of the source image in a JPEG encoder and decoder is divided into non-overlapping blocks of 8-by-8 pixels. Each block is then transformed using a two-dimensional discrete cosine transform ("DCT") with an 8-by-8 kernel.

The resulting 64 coefficients, computed at a 2-D array of 8-by-8 numbers, represent the frequency contents of the given block. The DCT coefficient, in the upper left hand corner of the 2-D array, measures the energy of the zero- frequency or ("DC") term. The remaining entries are ("AC") coefficients, which provide relative strengths of signal terms with increasing horizontal frequency from left to right, and for terms with increasing vertical frequency from top to bottom.

In subsequent steps, the DCT coefficients are quantized, reordered into an ascending frequency array, and encoded. This produces a data set associated with a particular image, which is substantially more compact than that associated with the bit-mapped image data from which it was obtained. Compressed data is expanded via implementation of an inverse algorithm to form a bit-mapped image corresponding to the earlier image. This image is suitably reproduced on a conventional video display terminal.

The present invention provides a new and improved system for video inspection, which utilizes data in its compressed form to accomplish high-speed inspection with improved speed, accuracy, and reduced hardware requirements.

Brief Summary of the Invention

The present invention contemplates a new and improved system for video inspection, which provides increased throughput inspection integrity, and may advantageously adapt existing data compression methodology. This new use is beyond the methods found in "Interesting Properties of the Discrete Cosine Transforms", Journal of Visual Communication and Image Representation, Vol. 3, No. 1, March, pp. 73-83, 1992. In accordance with the present invention, there is provided a means for acquiring compressed digitized image data, formed from a bit-mapped graphic image. This compressed data is compared to earlier acquired data representative of a flaw or aberration in a specimen under test. A comparison between these data allows for determining acceptability of the specimen.

In accordance with another aspect of the present invention, a means is provided for accomplishing image capture and real-time data compression of the digitized image data.

In accordance with yet another aspect of the present invention, there is provided a means for supplying more conventional bit-mapped inspection algorithms to bit¬ mapped data associated with compressed data, from which a determination of acceptability cannot be made.

In accordance with yet a further aspect of the present invention, data associated with analyzed Specimens is adapted to be archived, in compressed or uncompressed format, to provide an audit trail and acquire feedback information in accordance with manufacturing methods.

An advantage of the present system is the provision for a high-speed video inspection system that has lower processing power requirements. Another advantage of the present invention is the provision for a system of high speed inspection which utilizes commercially available conventional data compression technology.

Yet another advantage of the present invention is that by combining the functionality of an image compression system and an automated inspection system, the automated inspection system gains thereby all the functionality of an image compression system, including the ability to archive image data in a highly compressed form. Further advantages will become apparent to one of ordinary skill in the art, upon a reading and understanding of the subject specification.

Brief Description of the Drawings

The invention may take physical form in certain parts and arrangements of parts, a preferred embodiment of which will be described in detail in this specification and illustrated in the accompanying drawings which form a part hereof, and wherein:

FIGURE 1 illustrates, in block diagram form, an inspection system employing analysis of compressed image data in accordance with the present invention;

FIGURE 2 illustrates, in block diagram form, a JPEG image capture and analysis system for flaw determination in accordance with the subject invention; FIGURE 3 illustrates, in block diagram form, the preferred embodiment hardware implementation of the data compression subsection of the subject system;

FIGURE 4 illustrates, in block diagram form, a multi-processor system employed in a preferred embodiment of the subject system; and

FIGURE 5 illustrates, in block diagram form, an MPEG image capture and analysis system for flaw determination in accordance with the subject invention.

Detailed Description of The Preferred Embodiment

The subject application teaches the use of digitized image data, that is first transformed into the frequency domain, then compressed, to accomplish analysis of the data as it existed prior to transformation and compression. As used herein, "image" is defined as a data signature representative of physical attributes. Accordingly, an image includes, in addition to video or picture-based images, x-ray images, acoustic images, holographic images as well as any real-time data signal. The presently available compression technology, referred to above, employs a discrete cosine transform to achieve its compression. The DCT is advantageously employed in image compression insofar as it has been found to minimize artifacts and maximize the integrity of boundary or transitional image areas. The system may also suitably employ various frequency domain transforms.

A basic transform may be summed into the following equation:

In this basic equation, F(s) represents the transform of the earlier function f(t). The function K(st) is referred to as a kernel of the transformation. The kernel is the basic distinguishing factor between various of the transforms. The basic cosine transform is as follows:

It will be noted that for this cosine transform, the kernel is formed of cos(nπX/L).

All transforms also have an inverse. In the case of the basic cosine transform, the inverse is as follows:

$ [F_D] -fix) =-^+∑ ς-cos (nt-x/L)

J.-1

The cosine transform is best accomplished to a digitized data stream by implementation via the discretized form. This is given by:

BD(u, v)

Its inverse transform being:

. 7 7 bdix.y) =^∑∑ Ciu) Civ) BD(u, v) πu(2x+l) cos πv(2y+l)

U-O V-0 16 16

wherein: u ^~~ horizontal frequency index

v ≡ vertical frequency index

x ≡ horizontal posi tion index

y ≡ vertical posi tion index

C(i) a -- for i=0; 1 for i=l through 7

This DCT rendering is that achieved by present JPEG hardware realizations.

Although the discrete cosine transform is provided in the preferred embodiment, so as to maintain compliance with JPEG baseline standards, a host of other transformation could be applied in its stead. The DCT is a member of a class of linear transformations which serve to represent spatially correlated data with a less correlated data set. The transformation which is considered optimal for the representation of continuous tone image partitions with an uncorrelated data set is the Karhunen-Loeve transformation ("KLT") . However, no computationally efficient algorithms for the computation of the KLT have been developed so that transformations which approximate the performance of the KLT and for which fast algorithms exist, such as the DCT, are generally used in its stead. Other asymptotic equivalents to the KLT, which have been developed for use in image compression systems, include the Phase Shift Cosine Transform ("PSCT") and the Weighted Cosine Transform ("WCT") . Other transformations that could be used include the Fast Fourier Transform ("FFT") and the Hadamard Transform ("HT") , although these transformation are generally recognized as exhibiting inferior performance. Included within the scope of the present invention are provisions for use of any of the aforementioned transformations or any other transformation which serves the end of transforming spatially correlated data into a less correlated representation.

Referring particularly to the drawings wherein the showings are for the purpose of illustrating the preferred embodiment of the invention only, and not for the purpose of limiting the same, FIGURE 1 illustrates an inspection system A adapted to inspect a specimen stream B suitably formed of a plurality of discrete specimens 10 or a continuous or webbed specimen 12. The specimens are communicated to a viewing area 14 of a lens 16 and video camera or capture means 18 which functions as a data acquisition means. In the preferred embodiment, the video camera 18 is formed with a charge coupled device ("CCD") array to provide a bit-mapped pixel-based (spatial) captured image. A strobe control unit 24 synchronizes image capture of the video camera 18 with an illumination device 26. The illumination device 26 is advantageously comprised of an array of light-emitting diodes ("LEDs") in the preferred embodiment when still or "frozen" images of discrete specimens such as specimens 10 are being analyzed. The strobe control 24 advantageously allows the freezing of a specimen which is in motion relative to tfca video camera 18 while disposed in the viewing area 14.

An image captured by the CCD array of video camera 18 is accordingly formed as a bit-mapped (spatial) image. Each pixel of the bit-mapped image is assigned a gray-scale value, 0-255 in the preferred embodiment. This image is communicated to a data compression system or means 30 to accomplish the data compression transform. The particulars of suitable compression hardware will be detailed below.

The implementation of FIGURE 1 facilitates DCT- based compression suitable for use with JPEG and MPEG schemes. Compressed data is communicated to a defect detection system ("DDS") 32, the particulars of which will also be described in detail below. The DDS 32 communicates a signal DEFECT DETECT representative of acceptability of a specimen, or portion thereof in the case of continuous media, to the extent this is discernable therefrom. ^* All data representative of defects detected, or some representative portion thereof if this is deemed sufficient for classification and display purposes, is written to a buffer 36. The defect classification and display module (DCDM) 34 will be passed data representing the location of defect data within the buffer 36. The DCDM 34 will retrieve defect data from buffer 36 and perform further processing for the purpose of classification and display. Secondary storage 38 will be provided to permit long term archival of defect information.

Also illustrated is an input/output ("I/O") terminal 40 inclusive of a video display terminal ("VDT") or monitor. Incorporation of a VDT advantageously allows for selective viewing of the specimens, either in real-time as the inspections are completed, or of stored data, such as image data stored on secondary storage 38.

Turning now to FIGURE 2, the data compression subsystem 30, and a DDS 32, are illustrated in greater detail in a JPEG environment. This system illustrates the use of multiple cameras 18(1) through 18(n) . The images captured by each camera 18 are bit-mapped or pixel based and accordingly, spatial in nature. Data from camera system 18 is communicated to a raster block converter 44 and input lookup table 48 for gamma correction.

The data compression, suitably JPEG, is accomplished by dedicated hardware in the data compression subsystem 30. The hardware forming the compression subsystem of the preferred embodiment includes a DCT processor portion

30(a) comprised of an LSI DCT processor L64735 of LSI Logic

Corp., Milpitis, California. The DCT processor 30(a) works in concert with an JPEG coder 30(b) formed of an LSI L64745.

Compressed data from one or more compression subsystems 30 is communicated to the defect detection system 32. These data streams are each processed through a block anomaly database 46 which has been suitably constructed so as to recognize and categorize anomalous events occurring within a single block or stream of consecutive blocks. The database 46 may be suitably constructed as a memory device with sequential state feedback or as a software emulation of the former. Block anomaly reports generated by the block anomaly databases 46 are correlated with previous anomaly reports from the same database and/or from other databases, as applicable, in the data correlation subsystem 48, generating macro anomaly reports describing the properties of the anomaly as a whole. The macro properties of each anomaly are then compared against preselected defect discrimination criterion in defect discrimination subsystem 50. If an anomaly is determined, based on the selected defect discrimination criterion, to be a defect the data representative of this defect is fetched from data buffer 52 stored in buffer 36 along with the macro anomaly report itself, and the location of these reports within buffer 36 are reported to the DCDM 34. The DCDM 34 will in turn retrieve data from buffer 36 and perform additional processing as necessary for classification and display.

Turning now to FIGURE 3, illustrated in a block diagram is the system architecture of the preferred embodiment to accomplish the above-described function. The preferred embodiment employs a VME bus interconnection. In the illustration of FIGURE 3, the video capture means 18 is formed as a series of eight cameras. Cameras 18(a) - 18(d) feed a first video processor 74(a) and first video digitizer 76(a) of a data compression portion 30(a). Similarly, cameras 18(e) - 18(h) feed a second video processor 74(b) ?nd second video digitizer 76(b) of a second data compresε _ portion 30(b) . Signals thus digitized are provided to f ?t and second compression hardware processors 78(a) and lb _\,D) associated with data compression portions 30(a) and 30(b), respectively. As can be seen from the illustration, the video processors 74, video digitizers 76, and compression hardware processors 78 are all connected to VME'bus 80.

Also connected to the bus 80 is a specimen velocity sensor 82 which receives data representative of specimen velocity from an encoder 84. The encoder is suitably adapted for either discrete components, or continuous or web components, evidenced by FIGURE 1 in a fashion that will be appreciated by one of ordinary skill in the art. The velocity sensor is interconnected with the lighting control 24 which is interconnected with an input output ("I/O") unit 88. The I/O unit 88 is associated with a remote processor 90 which is, in turn, in data communication with bus 80. The I/O unit 88 therefore provides a means by which data communication may be accomplished by the entire system and an external device.

Also illustrated in FIGURE 3 is a camera alignment video monitor 92 which allows for feedback-control alignment of the video camera system 18. A video selector control 94 advantageously facilitates user-definable video parameters of the system.

Turning now to FIGURE 4 with added reference to FIGURE 3, additional description is made of the preferred multi-processor hardware platform. Disclosed generally at 70 is a master node processor. Disclosed at 72(a) is a representative SWP processor node, of which a plurality are advantageously provided. A second SWP processor is illustrated at 72(b) . Additional SWP processors are formed in a similar fashion to those 72(a) and 72(b). The master processor 70 is advantageously formed of a relatively high performance user interface engine 100 such as that achieved today by a reduced instruction set computer ("RISC") unit. In the preferred embodiment, this RISC engine is formed from a SPARC architecture. Also included is I/O engine 104, and host processor 118. The master processor 70, as well as the one or more SWP processor modules 72 are interconnected via a bus system 80', also formed as a VME bus in the preferred "embodiment. Several additional local busses are also supplied. A SCSI bus 102 provides a data path between master processor 70 and SWP processor module or modules 72 via I/O engine 104. An ETHERNET connection 96 advantageously provides a remote link between I/O engine 104 and user interface engine 100 of master processor 70. The master processor 70 is advantageously provided with a host processor board 118 which is, in turn, interfaced with an high bandwidth I/O interface as illustrated by I/O subsystem 120.

Several peripherals are advantageously utilized with the subject inspection system. Included therein is a network interface 130; a system console 132, inclusive of a keyboard subsystem and video display; a system hard disk 134; a defect display monitor 136; and a bulk storage medium such as hard disk 138. The system hard disk 134, the bulk storage hard disk 138 and a first storage hard disk 139 collectively form the disk subsystem 36 (FIGURE 1) . The first storage hard disk facilitates real-time capture and short-term storage of captured image data. It functions as a buffer for storage of data which may face a communication bottleneck such as might be attributed to ETHERNET connection 96. The system console 132 and the defect display monitor 136 collectively form the I/O unit 40 (FIGURE 1) .

In the preferred embodiment, the network interface 130 facilitates interconnection to plant I/O via an ETHERNET interface. The system hard disk 134, system console 132, and network interface 130 are all in data communication through a SCSI bus subsystem to user interface engine 100. Use of a SCSI bus system allows for adaptation of a substantial number of readily available peripherals.

The above-described defect display monitor 136 is advantageously placed in direct communication with user interface engine 100. The defect display monitor 136 allows for real-time operator viewing of images associated with specimens for which defects had been determined.^* The image storage hard disk 139 provides the non-volatile means for selective storage or archiving of defect data as described earlier.

Turning to the representative SWP processor modules 72(a) and 72(b) illustrated in FIGURE 4, data communication with the master processor 70 is accommodated through the bus 80' and the SCSI bus 102. For each SWP processor there is a corresponding high-speed receiver 142. Communication from the respective high-speed receiver 142 is made to its associated SWP processor 140 via an associated software processor bus 144. Each high-speed optical receiver 142 also receives data from compression hardware processors (FIGURE 3).

In the preferred embodiment, the SWP processor 72 utilizes five TMS320C31 DSPs on a single slot 6U VME card. The five DSPs on each SWP are organized as four slave DSPs with a single master DSP. The master DSP handles distribution of incoming data packets to its slaves. The slave DSPs, in turn, process their assigned data packets and report to the master DSP. The master DSP then performs inter-packet connectivity and rudimentary anomaly classification. When a defect, or an anomaly strongly suspected to be a defect is found, the master DSP will forward a compressed representation of the defect to the user interface engine via SCSI bus 102 and I/O engine 104 for further classification and display. Turning now to FIGURE 5, illustrated is an inspection system analogous to the inspection system illustrated in FIGURE 2, but employing a modified MPEG (Motion Picture Expert Group) compression system in place of the JPEG compression system. A. first distinction of the MPEG embodiment is provided in connection with the MPEG compression unit 180. The unit 180 facilitates time frequency domain transfer information to accomplish the MPEG compression. The components forming the MPEG compression unit 180 are commercially available components of LSI Logic Corp. A quantization processor portion 182 is formed within an L64740 chip. An interframe processor portion 184 is suitably formed from an L64760 chip. A discrete cosine transform operation is completed by a DCT transform chip 186 suitably formed from an L64730 DCT processor. The VLC 188 is suitably formed from an L64750 variable length coder. The BCH 190 is suitably formed from an L64715 BCH coder. A motion estimation processor 192 is suitably formed from an L64720 chip.

MPEG systems are designed to compress a sequence of similar images, such as typical frames of motion picture video data, using interframe comparisons. In a conventional MPEG system the comparison data supplied to the interframe processor 184 would be generated from the previous frame data, however, since the present invention requires the compression not of motion video but of images associated with a sequence of similar specimen or frames of a continuous media with a repeating pattern, the present invention contemplates the use of a modified MPEG compression system which supplies the interframe processor 184 with image data of a reference frame against which subsequent frames would be compared.

The compression data generated by this modified MPEG system can be analyzed for the purpose of detecting and classifying defects in a manner exactly analogous to that used in the JPEG compression based system illustrated in FIGURE 2, although it would be ε predated that the data stored in the databases 46' may di. er markedly from that in databases 46 due to differences in the MPEG data format.

A decision to update compression ratios, as determined at block 52', causes a signal to the quantization process 182 of MPEG compression unit 180. This signal dictates an update of quantization tables provided therein. The invention has been described with Reference to the preferred embodiment. Obviously, modifications and alterations will occur to others upon the reading and understanding of this specification. It is intended to include all such modifications and alterations insofar as they come within the scope of the appended claims or the equivalents thereof.

Claims

Having thus described our invention, we now claim:

1. An industrial inspection system comprising: data acquisition means for acquiring compressed digitized image data formed from domain transformed spatially encoded data of an associated image; detection means for detecting anomalies in a specimen represented in said image; storage means for storing data representative of a plurality of aberration pattern types; comparison means for selectively classifying anomalies detected by the detection means as being of a type represented in the storage means; and test means for generating a test signal representative of acceptability of a specimen associated with the image in accordance with output of the comparison means.

2. The industrial inspection system of claim 1 further comprising data compression means for forming the compressed image data from an image communicated thereto.

3. The industrial inspection system of claim 2 wherein the detection means is comprised of: a means of detecting aberration constituents in individual domains in said domain transformed spatially encoded data; and a means of selectively associating subsets of said aberration constituents with anomalies.

4. The industrial inspection system of claim 3 further comprising a means of selectively merging at least one set of aberration constituents associated with a first image with at least one set of aberration constituents associated with a second image.

5. The industrial inspection system of claim 4 wherein the storage means represents aberration pattern types by data structures which permit the unique differentiation of all aberration pattern types based on the properties stored in the associated data structures.

6. The industrial inspection system of claim 5 wherein the comparison means is comprised of: a means of extracting macro anomaly properties from the set of aberration constituents associated with an anomaly by the detection means; and a means of classifying an anomaly as being of one of the aberration pattern types defined by the storage means, by selectively comparing the macro properties extracted therefrom with the properties of the aberration pattern types defined in the data structures stored by the storage means.

7. The industrial inspection system of claim 6: further comprisingmeans for communicating the data to the data compression means as a series of bit-mapped data frames; and wherein the data compression means further includes means for forming the compressed image data as a series of compressed image data segments corresponding to the series of bit-mapped data frames communicated thereto.

8. The industrial inspection system of claim 7 wherein the data compression means further includes means for forming the compressed image data by application of a cosine transformation.

9. The industrial inspection system of claim 8 further comprising feedback means for communicating feedback data to the compression means in accordance with an output of the comparison means.

10. The industrial inspection system 'of claim 9 further comprising: a video camera for capturing the image data; an illumination system adapted to illuminate an associated specimen disposed in a field of view of the video camera; and means for synchronizing operation of the video camera with the illumination system.

11. The industrial inspection system of claim 10 further comprising: data decompression means for selectively decompressing the series of compressed image data segments to form decompressed spatially-encoded data frames corresponding thereto.

12. The industrial inspection system of claim 11 further comprising means for selectively storing in a non¬ volatile data storage device at least one of the compressed digitized image data and the spatially encoded image data associated therewith.

13. The industrial inspection system of claim 12 further comprising means for activating the data compression means in accordance with an output of the test means.

14. The industrial inspection system of claim 13 further comprising means for selectively archiving the compressed digitized image data.

15. The industrial inspection system of claim 14 further comprising means for cataloging archived compressed image data in accordance with a test signal associated therewith.

16. A method of fault detection utilising a data processor comprising the steps of:

(a) receiving compressed digitized domain transformed data; (b) storing the compressed digitized domain transformed data in an associated data storage device as stored compressed digitized data;

(c) processing compressed digitized data so as to extract therefrom aberration constituents associated with individual transform domains;

(d) selectively associating subsets of said aberration constituents with anomalies, based on predefined inter-domain association criterion;

(e) extracting a set of macro anomaly properties from the set of aberration constituents associated with an anomaly;

(f) classifying an anomaly as being on at least one of a set of predefined aberration pattern types, by selectively comparing the macro properties extracted therefrom with the properties of the predefined aberration pattern types; and

(g) generating a signal representative of acceptability of a specimen associated with the compressed digitized image data in accordance with the anomaly properties generated in step (e) as classified in step (f) .

17. The method of claim 16 further comprising the steps of: generating a test signal representative of a failure to determine acceptability of the specimen in accordance with step (g) ; selectively decompressing the compressed digitized domain transformed data into spatially encoded data .

18. The method of claim 17 further comprising the steps of: storing the spatially encoded data in an associated data storage device as stored spatially encoded data; accessing spatially encoded comparison data representative of at least one preselected secondary aberration constituent; selectively comparing, in the data processor, the stored spatially encoded data to the spatially encoded comparison data; and generating a secondary test signal representative of acceptability of a specimen associated with the compressed digitized domain transformed data in accordance with an output of the comparison means.

19. The method of claim 18 further comprising the step of selectively communicating at least one of the compressed digitized domain transformed data and the spatially encoded data to at least one of an archival memory and a video display terminal.

20. The method of claim 19 wherein step (a) further includes the step of receiving the compressed digitized domain transformed data as a series of compressed image data segments corresponding to a series of spatially encoded data segments, and wherein the step of selectively decompressing includes the step of selectively generating spatially encoded data segments corresponding to compressed image data segments corresponding thereto.

21. An industrial inspection system comprising: video capture means for capturing a series of bit-mapped image data frames; data compression means for forming a series of domain transformed image data frames corresponding to each of the bit-mapped graphic image data frames communicated thereto from the video capture means; detection means for detecting anomalies in specimen represented in said bit-mapped graphic image; storage means for storing data representative of a plurality of aberration pattern types; comparisonmeans for classi ying the anomalies detected by the detection means as being of a type represented in the storage means; test means including means for generating a first test signal representative of acceptability of a specimen associated with a given domain transformed image data frame in accordance with an output of the comparison means, the test means including means for generating a second test signal representative of a failure to determine acceptability of the specimen in accordance with an output of the comparison means; the comparison means including means for selectively comparing, in accordance with the second test signal, a bit-mapped graphic image data frame associated with the specimen with bit-mapped comparison data representative of acceptability of the specimen; and the test means further including means for generating a third test signal representative of acceptability of the specimen associated with the bit-mapped graphic image in accordance with an output of the comparison means.

22. The industrial inspection system of claim 21 wherein the data compression means includes means for forming domain transformed image data frames in accordance with at least one of MPEG and JPEG data compression formats.

23. The industrial inspection system of claim 21 wherein the compression means includes means for generating

MPEG format compressed data wherein interframe comparisons are performed on each frame as compared to an absolute reference frame.

24. The industrial inspection system of claim 23 further comprising means for selecting from a set of possible absolute reference frames based on a prior knowledge of the specimen represented in at least one of the image frame and knowledge derived from preprocessing the image frame.

25. The industrial inspection system of claim 22 further comprising: means for communicating a series of specimens to a viewing area associated with video capture means; illumination means adapted for illuminating a specimen disposed in the viewing area; and means for selectively enabling the illumination means.

26. The industrial inspection system of claim 21 further comprising means for selectively archiving the domain transformed image data in accordance with the first test signal.

AMENDED CLAIMS

[received by the International Bureau on 20 December 1993 (20.12.93); original claims 1,16 and 21 amended; other claims unchanged (3 pages)]

1. An industrial inspection system comprising: data acquisition means for acquiring compressed digitized image data which has been reversibly compressed by application of a domain transform to spatially encoded data of an associated image; detection means for detecting anomalies in a specimen represented in said image; storage means for storing data representative of a plurality of aberration pattern types; comparison means for selectively classifying anomalies detected by the detection means as being of a type represented in the storage means; and test means for generating a test signal representative of acceptability of a specimen associated with the image in accordance with output of the comparison means.

4. The industrial inspection system of claim 3 further comprising a means of selectively merging at least one set of aberration constituents associated with a first image with at least one set of aberration constituents associated with a second image. therewith.

16. A method of fault detection utilizing a data processor comprising the steps of:

(a) receiving compressed digitized domain transformed data which has been reversibly compressed by application of a domain transform;

(b) storing the compressed digitized domain transformed data in an associated data storage device as stored compressed digitized data;

(d) selectively associating subsets of said aberration constituents with anomalies, based on predefined inter-domain association criterion; (e) extracting a set of macro anomaly properties from the set of aberration constituents associated with an anomaly;

17. The method of claim 16 further comprising the steps of: generating a test signal representative of a failure to determine acceptability of the specimen in accordance with step (g) ; selectively decompressing the compressed digitized domain transformed data into spatially encoded bit-mapped image data frames; data compression means for forming a series of reversible, domain transform compressed image data frames corresponding to each of the bit-mapped graphic image data frames communicated thereto from the video capture means; detection means for detecting anomalies in specimen represented in said bit-mapped graphic image; storage means for storing data representative of a plurality of aberration pattern types; comparisonmeans for classifying the anomalies detected by the detection means as being of a type represented in the storage means; test means including means for generating a first test signal representative of acceptability of a specimen associated with a given domain transformed image data frame in accordance with an output of the comparison means, the test means including means for generating a second test signal representative of a failure to determine acceptability of the specimen in accordance with an output of the comparison means; the comparison means including means for selectively comparing, in accordance with the second test signal, a bit-mapped graphic image data frame associated with the specimen with bit-mapped comparison data representative of acceptability of the specimen; and the test means further including means for generating a third test signal representative of acceptability of the specimen associated with the bit-mapped graphic image in accordance with an output of the comparison means.