CN103234472A - Detecting method and detecting system for fiber fineness and density of Rex-rabbit clothing hair - Google Patents

Abstract

The invention provides a detecting method and a detecting system for fiber fineness and density of Rex-rabbit clothing hair. The detecting method includes a, collecting clothing hair image data, b, transmitting the clothing hair image data to a computer to be stored, and c, completing detection analysis through a computer processing software. The detecting system comprises a clothing hair image data collection device and a clothing hair image data storage and detection analyzing device. By the detecting method, Rex-rabbit fur quality, clothing hair density and density results of the clothing hair acquired from a living Rex rabbit can be detected and analyzed directly, so that the detecting method is convenient to operate and high in efficiency; and the detecting system is simple in structure, convenient to carry, and high in detecting speed and accuracy.

Description

A kind of otter rabbit is by wool fibre fineness, Density Detection method and system

Technical field

The invention belongs to the precision measuring instrument technical field, be particularly related to the diameter measurement method and apparatus of the textile fibres of fur-bearing animal pelage qualities such as always comprising the rabbit hair, wool, ermine suede, mohair yarn and other round sections, be widely used in the quality management of fur product, textile import and export and production industry.

Background technology

The otter rabbit has another name called makes every effort to overcome this rabbit (Rex rabbit), former meaning " king in the rabbit ".The Yin Likesi rabbit is smooth upright by hair again, is rich in gorgeous gloss, and soft, comfortable, capillary is close, very like precious fur-bearing animal otter.So, how to be referred to as with " otter rabbit ".The otter rabbit is that typical skin is used rabbit.The requirement of otter coney quality standard is summarised as " short, thin, close, flat, beautiful, jail ".So-called " weak point " is exactly that wool fibre is short, and wool fibre length is between 1.3 centimetres～2.2 centimetres." carefully " be exactly the wool fibre diameter at 16 microns～18 microns, coarse wool is few." close " is exactly that fine hair is abundant, by gross density at 10000～25000/cm ²" put down " and be exactly fine hair length unanimity, smooth." U.S. " is exactly that tone is attractive in appearance, and be glossy." jail " be exactly by hair give birth to firmly difficult drop-off.The beaver rabbit skin hair is sold always in the international market and is expected.

Relate to an a kind of quilt hair quality detection apparatus (100) 2007.04.11 notification number is the patented claim of CN1946335, it has electromagnetic radiation source (80) and imaging sensor (74), and has radiation selecting arrangement (83,48).In the use of this equipment, this selecting arrangement improve emission be linked into skin (8), in skin via scattering repeatedly and homogenising and not absorbing, and arrive this sensor (74) and that part of radiation of skin image is provided, and for example arrive ratio between that part of radiation of sensor in the skin surface reflection via alternate manner.Select by means of this, the contrast of image can be improved, and can still less depend on skin color and skin pseudomorphism, thereby can make the white on test example such as the light skin easier by hair.This is mainly used in the checkout equipment of live body fur-bearing animal pelage quality by the hair checkout equipment, existing otter rabbit is by capillary degree, Density Detection method, mainly contain artificial cropping counting method, microscope sciagraphy and get the skin section statining and detect hair follicle fiber counting method, need live body to get skin, cropping on the one hand, damage fur or butcher the otter rabbit; This detecting instrument complex structure on the other hand, detection can only be at the experiment in-house operation, and the workload large period is long, and be not easy to data and handle, the labor intensive material resources, detection cost height has been not easy to the quality management of production industry.

Summary of the invention

Introduced the concept of a series of reduced forms in the summary of the invention part, this will further describe in the embodiment part.Summary of the invention part of the present invention does not also mean that key feature and the essential features that will attempt to limit technical scheme required for protection, does not more mean that the protection domain of attempting to determine technical scheme required for protection.

One of the object of the invention provides a kind of otter rabbit by wool fibre fineness, Density Detection method, can get galley proof live body otter rabbit and directly detect analysis, and is easy to operate, efficient is high.

The present invention provides a kind of otter rabbit by wool fibre fineness, Density Detection system simultaneously, and it is simple in structure, be convenient for carrying, and detection speed is fast, precision is high.

The present invention is achieved through the following technical solutions,

A kind of otter rabbit be is characterized in that comprising the steps: by capillary degree, density fiber detection method

A, collection are by a hair view data;

B, will in computing machine, be stored by the hair image data transmission;

C, finish by computer-processing software and detect to analyze.

The method according to this invention adopts detector directly to gather by the hair view data from live body otter rabbit on one's body.

The method according to this invention adopts the portable data treatment facility to finish the Data Detection analysis in step c.

Detector according to the present invention is specially DinoLite hand-held USB digit microscope.

The otter rabbit be is characterized in that comprising by hair image data acquiring device, by hair image data storage and detection analytical equipment by wool fibre fineness, Density Detection system.

According to system of the present invention, described image data acquiring device is DinoLite hand-held USB digit microscope.

According to system of the present invention, described is the portable data treatment facility by hair image data storage and detection analytical equipment.

According to system of the present invention, described microscope mainly is made of CCD, convex lens and lighting unit.

According to system of the present invention, more convenient for using, described portable data treatment facility arranges promising input/output interface and charge port.

The present invention can get galley proof live body otter rabbit and directly detect analysis, and simple in structure, be convenient for carrying, detection speed is fast, precision is high, uses easy to operate.Particularly, have following 4 outstanding features:

1, quick: namely survey namely, analysis of image data is added up, and obtains the detection data result of this otter rabbit rapidly, and storage is got conveniently, but also ex-post analysis, output.

2, portable: instrument is designed to hand-held, and is easy and simple to handle, need not alternating current, is fit to the field, the on-the-spot detection.

3, harmless: directly detect at living animal, harmless to detected fur-bearing animal fur, repeatedly duplicate measurements.

4, change different sampling heads, can detect multiple animal skin.

Description of drawings

Fig. 1 is the structural representation of detection system among the present invention.

Fig. 2 is the BMP format-pattern of 640X480.

Fig. 3 is undressed picture.

Fig. 4 is the picture after gray scale is handled.

The color range spectrogram of Fig. 5 for Fig. 4 is done.

Fig. 6 is color range spectrogram exemplary plot.

Fig. 7 is the Pixel Information synoptic diagram of image relevant position.

Fig. 8 a and Fig. 8 b are respectively two-dimensional adaptive denoising filtration treatment exemplary plot.

Fig. 9 a and Fig. 9 b are respectively and create predefine filtrator exemplary plot.

Figure 10 a and 10b are respectively the border exemplary plot in the identification intensity image.

The nothing intersection synoptic diagram that Figure 11 is presented at microscopically by hair for the otter rabbit.

Intersection or overlapping synoptic diagram that Figure 12 is presented at microscopically by hair for the otter rabbit.

Figure 13 for from left to right image is carried out sectioning gained synoptic diagram.

Figure 14 is data-gray-value variation synoptic diagram.

Figure 15 is the statistics synoptic diagram.

Figure 16 is denoise algorithm figure.

Wherein 1 is that a quilt mao image data acquiring device, 2 is that quilt mao image data storage and detection analytical equipment, 3 are that CCD, 4 is that convex lens, 5 are that lighting unit, 6 is that input/output interface, 7 is charge port.

Embodiment

In the following description, a large amount of concrete details have been provided in order to more thorough understanding of the invention is provided.Yet, it will be apparent to one skilled in the art that the present invention can need not one or more these details and implemented.In other example, for fear of obscuring with the present invention, be not described for technical characterictics more well known in the art.

In order thoroughly to understand the present invention, detailed details will be proposed in following description so that illustrate the present invention be how to solve existing communication perception evaluating system can't collect problems such as classification to the network problem zone.Obviously, execution of the present invention is not limited to the specific details that the technician of the communications field has the knack of.Preferred embodiment of the present invention is described in detail as follows, yet except these were described in detail, the present invention can also have other embodiments.

A kind of otter rabbit be is characterized in that comprising the steps: by wool fibre fineness, Density Detection method

A, collection are by a hair view data;

B, will in computing machine, be stored by the hair image data transmission;

C, finish by computer-processing software and detect to analyze.

Adopt DinoLite hand-held USB digit microscope directly to gather by the hair view data from live body otter rabbit among the present invention on one's body.In step c, adopt the portable data treatment facility to finish the Data Detection analysis.

The present invention also provides a kind of otter rabbit by wool fibre fineness, Density Detection system simultaneously, it is characterized in that comprising by hair image data acquiring device 1, by hair image data storage and detection analytical equipment 2.

Further, image data acquiring device of the present invention is DinoLite hand-held USB digit microscope.

Further, of the present invention is the portable data treatment facility by hair image data storage and detection analytical equipment 2.

Described DinoLite hand-held USB digit microscope mainly is made of CCD3, convex lens 4 and lighting unit 5.

More convenient in order to use, described portable data treatment facility arranges promising input/output interface 6 and charge port 7.

The otter rabbit is as follows substantially by the software function of hair detector: a) store the otter rabbit by the hair image by the video heads collection.B) but the otter rabbit is formed the form of analytical calculation by the hair image.C) required the index Design mathematical model by the otter rabbit by the hair detection.D) required design result statistics interface according to the otter rabbit by the hair examining report.

The otter rabbit is by the collection of the micro-amplified video head of USB2.0 independence hand-held by the image acquisition of hair detector.The image acquisition part that adopts the programming of Directshow technology that USB device (micro-amplified video head) is write this software.

View data is divided into 2 classes: dynamic video image (avi file), static image file (bmp or jpg form).These 2 class data that obtain after catching are got static image file (bmp or jpg form) as analyzing data.Adopt the BMP form as the final analysis data.

Gather the BMP format-pattern of 640X480, from the lower left corner, finish in the upper right corner with the BMP picture, altogether 640X480 point.PBits[640 of reputation] array of a RGB class of [480], the data that each matrix dot obtains are rgb values, the RGB class be one by R, G, B data are formed, and represent the red R of this point respectively, green G, the numerical value of blue B three-primary colours.As shown in Figure 2, can see just this 640X480 the point data rgb value that can access.Because the data that need to handle are white or brown, the hair of black is handled so need carry out gray scale to data.Processing procedure is as follows:

The classical operational formula of RGB gray-scale value process gray scale of each point obtains the data value of (0-255), has so just obtained a shaping two-dimensional array P[640] [480].The bright-dark degree that each point of these data has just reflected each position of picture.This array array is exactly a basis of back mathematical modeling.Have a look the result that gray scale is handled below.Fig. 3 is not for there being treated picture, and Fig. 4 is a picture that gray scale was handled, and can see the gray-scale value of position of hair and the difference of background color.

Data model analyzing and processing process

The just P[640 that obtains] the data array of [480].

The first step color range of handling is analyzed:

For example last figure is done the color range analysis, obtain color range spectrogram shown in Figure 5.Color range branch utilizes this color range partly to separate background color and useful data, and then the hair partial data is carried out analyzing and processing as seen from Figure 5.Certainly not being that each picture can both effectively be handled, is example with Fig. 6, can see when color range is defined as 109, and the hair part can not well show, and has also mixed background color inside.This explanation background color and hair color are too approaching, and hair color and background color intersect very serious in other words.In other words, for example the hair color range between (110-130), and ambient color, between the background color (80-150), this just makes useful data to reject out.Because the gray scale of carrying out is handled, so 3 yuan of looks have become the monobasic look, this process is that data are lost, but the gray scale processing is the method the simplest that image tonescale is rejected.So mathematical model can not simply be handled to cut apart with reference to gray scale, need carry out other method modeling from 3 yuan of color base plinth.For example to the R look, the G look, 3 kinds of colors of B look are all carried out the dividing processing refinement.But individual one very difficult problem is arranged here, and it is white, greyish white that the color of hair mostly is greatly, black, palm fibre, and the rgb value of these colors is close to 3 five equilibriums.Even if it is cut apart analysis so carry out 3 looks, actual the same with the gray scale segmentation effect.So needing data source can effectively distinguish hair and ambient color.This just requires to add background color and distinguishes.After imagination background color and hair color can effectively be distinguished, can obtain the data coordinates position of hair, this coordinate position point has been exactly corresponding P[640] array position (horizontal stroke, column position) of [480] this 2-D data.These positions have been arranged, used an effective and feasible data modeling again, can measure the diameter of hair, quantity, results such as unit area radical.

The hair radical calculates in fact, and this process is exactly so-called data gathering analysis.

It is that geometric shape calculating modeling analysis is added in the statistical computation of a data slope that the diameter of hair calculates.

4) foundation of data model

At first, the data structure of for the treatment of data.As shown in Figure 7, shaping two-dimensional array P[640] [480] comprised 640*480 point, each point (X, Y) Dui Ying value correspondence the Pixel Information of image relevant position.

Because the data of each point of image of handling no longer have been rgb values, but gray-scale value, the information of each point has only a value.Again and can be with P[640] [480] regard matrix array data as.When designing a model, it is that the basis is designed that this data array is arranged.Here need to use and also have a very important mathematical tool Matlab.Matlab can carry out a large amount of complex mathematical computations.Because can not meeting mathematical model fully, handles image.In order better to set up and to design a model, need blur denoising to view data.Picture color range conversion after treatment is more steady, and such data more are conducive to analyze.

Utilize the powerful view data processing capacity of Matlab earlier image to be carried out denoising: wiener2

Function:

Carry out two-dimensional adaptive denoising filtration treatment.

Grammer:

J = wiener2(I,[m n],noise)

[J,noise] = wiener2(I,[m n])

For example

I = imread('saturn.GIF');

J = imnoise(I,'gaussian',0,0.005);

K = wiener2(J,[5 5]);

imshow(J)

figure, imshow(K)

Shown in Fig. 8 a and Fig. 8 b, all processing operations all will be around the image data formatization that begins most to carry out, and the two-dimensional array after the datumization is operated.The otter rabbit is not too desirable by the hair image effect sometimes, when environmental noise is too big, also will do operations such as filtering processing.

freqspace

Function:

Determine the frequency space of two-dimentional frequency response.

Grammer:

[f1,f2] = freqspace(n)

[f1,f2] = freqspace([m n])

[x1,y1] = freqspace(...,'meshgrid')

f = freqspace(N)

f = freqspace(N,'whole')

Related command:

fsamp2, fwind1, fwind2

freqz2

Function:

Calculate two-dimentional frequency response.

Grammer:

[H,f1,f2] = freqz2(h,n1,n2)

[H,f1,f2] = freqz2(h,[n2 n1])

[H,f1,f2] = freqz2(h,f1,f2)

[H,f1,f2] = freqz2(h)

[...] = freqz2(h,...,[dx dy])

[...] = freqz2(h,...,dx)

freqz2(...)

For example

Hd = zeros(16,16);

Hd(5:12,5:12) = 1;

Hd(7:10,7:10) = 0;

h = fwind1(Hd,bartlett(16));

colormap(jet(64))

freqz2(h,[32 32]); axis ([–1 1 –1 1 0 1])

fsamp2

Function:

Design two-dimentional FIR filtrator with the frequency sampling method.

Grammer:

h = fsamp2(Hd)

h = fsamp2(f1,f2,Hd,[m n])

For example

[f1,f2] = freqspace(21,'meshgrid');

Hd = ones(21);

r = sqrt(f1.^2 + f2.^2);

Hd((r<0.1)|(r>0.5)) = 0;

colormap(jet(64))

mesh(f1,f2,Hd)

Related command:

conv2, filter2, freqspace, ftrans2, fwind1, fwind2

fspecial

Function:

Create the predefine filtrator.

Grammer:

h = fspecial(type)

h = fspecial(type,parameters)

For example

I = imread('saturn.GIF');

h = fspecial('unsharp',0.5);

I2 = filter2(h,I)/255;

imshow(I)

figure, imshow(I2)

Shown in Fig. 9 a and Fig. 9 b.

Related command:

conv2, edge, filter2, fsamp2, fwind1, fwind2

ftrans2

Function:

Design two-dimentional FIR filtrator by frequency inverted.

Grammer:

h = ftrans2(b,t)

h = ftrans2(b)

For example

colormap(jet(64))

b = remez(10,[0 0.05 0.15 0.55 0.65 1],[0 0 1 1 0 0]);

[H,w] = freqz(b,1,128,'whole');

plot(w/pi–1,fftshift(abs(H)))

Related command:

conv2, filter2, fsamp2, fwind1, fwind2

fwind1

Function:

Design two-dimentional FIR filtrator with the one dimension windowhood method.

Grammer:

h = fwind1(Hd,win)

h = fwind1(Hd,win1,win2)

h = fwind1(f1,f2,Hd,...)

For example

[f1,f2] = freqspace(21,'meshgrid');

Hd = ones(21);

r = sqrt(f1.^2 + f2.^2);

Hd((r<0.1)|(r>0.5)) = 0;

colormap(jet(64))

mesh(f1,f2,Hd)

Related command:

conv2, filter2, fsamp2, freqspace, ftrans2, fwind2

fwind2

Function:

Design two-dimentional FIR filtrator with two-dimentional windowhood method.

Grammer:

h = fwind2(Hd,win)

h = fwind2(f1,f2,Hd,win)

For example

[f1,f2] = freqspace(21,'meshgrid');

Hd = ones(21);

r = sqrt(f1.^2 + f2.^2);

Hd((r<0.1)|(r>0.5)) = 0;

colormap(jet(64))

mesh(f1,f2,Hd)

Related command:

conv2, filter2, fsamp2, freqspace, ftrans2, fwind1

Can be used for analyzing through view data after these complicated operations.Can use a very important operation this moment, and namely the edge is sought.edge

Function:

Border in the identification intensity image.

Grammer:

BW = edge(I,'sobel')

BW = edge(I,'sobel',thresh)

BW = edge(I,'sobel',thresh,direction)

[BW,thresh] = edge(I,'sobel',...)

BW = edge(I,'prewitt')

BW = edge(I,'prewitt',thresh)

BW = edge(I,'prewitt',thresh,direction)

[BW,thresh] = edge(I,'prewitt',...)

BW = edge(I,'roberts')

BW = edge(I,'roberts',thresh)

[BW,thresh] = edge(I,'roberts',...)

BW = edge(I,'log')

BW = edge(I,'log',thresh)

BW = edge(I,'log',thresh,sigma)

[BW,threshold] = edge(I,'log',...)

BW = edge(I,'zerocross',thresh,h)

[BW,thresh] = edge(I,'zerocross',...)

BW = edge(I,'canny')

BW = edge(I,'canny',thresh)

BW = edge(I,'canny',thresh,sigma)

[BW,threshold] = edge(I,'canny',...)

For example

I = imread('rice.GIF');

BW1 = edge(I,'prewitt');

BW2 = edge(I,'canny');

imshow(BW1);

figure, imshow(BW2)

Shown in Figure 10 a and Figure 10 b.

Utilize the interface of VC and Matlab, finish above image processing operations with a large amount of useful view data processing capacity that VC calls in the Matlab kernel.The data that obtain at last are exactly formally to set up the raw data of algorithm.Here he is defined as PCAL[640] [480] below will carry out modeling to computational data.

At first see this raw data prototype, as follows under the optimal situation of data prototype:

Situation one: as shown in figure 11, what all otter rabbits were presented at microscopically by hair is do not have to intersect.But this is impossible.Be like this under the ideal state

Situation two: as shown in figure 12, by hair cruces or overlapping, very general of this situation and appearance are a lot.

Situation three: not fogging clear by hair, edge fog.

Even if this situation has been passed through denoising, fuzzy, still can not well differentiate after the filtering, operations such as edge identification.

To sum up, even if with complicated human brain both artificially identification, also be difficult to pick out, when requiring on the one hand from the otter rabbit by the result of hair detector, allow by hair direction unanimity as much as possible, overlapping minimizing.On the other hand will be from the theory of probability statistics for coming out according to seeking early a feasible statistical method.

From the signal analysis and processing theory, design a feasible mathematical model algorithm, the calculating of adding probability statistics can reach the resultant error requirement that the otter rabbit is detected by hair basically.

The general signal of handling of signal analysis all is point-to-point data, the data of serializing.Suppose that all hairs all are the tops downward " growth " from image, this requirement can be accomplished by the structural adjustment of data acquisition head.Suppose all growths downwards from image top of all hairs so, certainly, if from left to right image is carried out sectioning, as shown in figure 13.

Can see red line the position of process, the conversion of pixel can be represented with a continuous serializing signal:

For example: black, black, black, white, white, white, black, black, black, white, white.The data of this serializing.

Certainly actual image can not obtain so clearly demarcated data by top operation.Gray-scale value is from 0-255, and 0 representative is extremely black, and 255 representatives are extremely white, provides one group of serializing real data afterwards here.

Come as can be seen from Figure 14 in fact, from the gradation of image value of certain delegation 640 data altogether from left to right, the variation of these data has just reflected the variation of gray-scale value.The data model of setting up just requires to identify the average gray of background color, is as the criterion with this value, defines a threshold line, and the value of this root threshold line just can be classified out by hair location of pixels partly by the location of pixels of hair part and Bush with data.So just can be with continuous partly being obtained by gross value, this continuous some length by the hair copy is exactly the quilt hair " false conjugate " of this section group the inside, so being defined as " false conjugate " most is because this diameter is not vertical with hair growth direction institute, need to revise, from here as can be seen, a statistics as shown in figure 15.

As shown above, if search out 4 groups of copies on threshold line in section group the inside, a continuous so as can be seen quilt hair color has partly comprised 4 groups, can think that these 4 groups is exactly the hair process radical of this sliced section.If a width of cloth figure is done N section, probability statistics are passed through in N section more so, can obtain a believable hair radical value, and the error of this value should be accepted within statistical error.

But again a kind of situation can appear here, hair intersects, overlapping, after 2 or 2 overlapping intersections of above hair, can be that the data volume of some hair groups the inside, single section group the inside substantially exceeds single by the normal pixel number of hair by the section group particularly.So also will design a mathematics manipulation formula overlapping hair is cut apart, the hair group number of single section group the inside is exactly the hair radical after cutting apart.The hair radical value of N section group is carried out, remove maximum, after the minimum, folded average value afterwards is exactly believable hair radical.

As for diameter of hair, the diameter of difference section group need be compared, design a slope group, the representative of this slope group be, each hair section group, the information that the hair collection location occurs forms a set group, generate a straight line formula, the slope of this straight line formula is exactly the modified value of diameter of hair, after the correction, can obtain the conjugata vera of hair.

Here describe the design of data threshold line and the method for statistic of classification in detail:

Know that from above in fact each section group is exactly the data sequence of each row of view data correspondence.If hair has passed this row, the part of having only hair to pass so is the color of hair, other all be background colour because background colour definition for black, even if there is illumination to disturb, also can not present the color of hair.So fine differentiation.

The trend of color so for example, beginning is background colour, begin mobile from left to right, when beginning quilt hair part to occur be, color begins to shoal, be exactly that gray-scale value begins to become big from little, when being greatly, be exactly by the shake of a position part after the edge of hair, but the shake of this numerical value is all within quilt hair color gamut to certain value, begin afterwards slowly to leave by the hair part, variation from diminishing greatly again enters the background colour part after leaving, afterwards circulation, need the data of statistics to be, when begin to occur by hair, when leave by the hair part, what this was kept is exactly by " the false length " of hair by the hair part, one total how many groups copy occurred by hair, are exactly by the radical of hair on this section group.

The array that defines this section group is B[640]

At first calculate the maximum gradation value Bmax of array, minimum gradation value Bmin is with BY=(Bmax+Bmin)/2 be defined as threshold line.Can design following threshold line Department of Statistics branch so.Owing to the slightly problem of shake of pixel can occur, also to design a threshold line dither noise disposal route, define color shake pixel value is no more than 5, and the value with 5 pixels about the BY quantizes so, if a certain pixel value is greater than BY-5, less than BY+5, so all this value is thought BY-5, these values are all the time less than BY+5, so as long as find the value that BY+5 is above so, can think that just pixel value has passed through threshold line BY, so just can remove shaking by a small margin.

The design of data statistics result report

The design of statistics report must be implemented according to the designing requirement of Gneral analysis software.Roughly need to comprise following several contents:

1) single image statistics, even click " collection " afterwards, image begins to gather, and needs afterwards present image is implemented to carry out analytic statistics.

2) after the single image statistics, must add standby database to, for example: after a plurality of positions of otter rabbit are added up, the several times data of these statistics need be integrated average treatment.

3) form form, histogram is convenient to the user and is observed.

4) can the result that every otter rabbit is measured be preserved, and can call at any time.

5) printing function because utilization is state-of-the-art UMPC embedded OS technology, so the palm portable system is supported the Windows printing function, so only need connect general printer, only need be write the printing data format that needs and get final product.

The whole software design cycle has been used DirectShow, and the programming of GDI image utilizes the VC++ programming tool to write software.

Several steps of the realization most critical of whole software are:

1, the foundation of mathematical model

The quality of mathematical model has directly influenced the confidence level of data result, multiplicity, even directly influenced and can measure the result.

2, the collection quality of data source

The fine or not front of so-called data source was also mentioned, and this process is the most difficult step outside data model is set up, and computer software design is good more after all, and it is intelligent again that model is set up, and data source is undesirable, also cannot realize measuring.The analytic function of computer programming design can not be compared with human brain, and perhaps the thing that can distinguish of people's naked eyes can not be realized by computing machine.So will be within the computer capacity scope design data model and collect the acquisition process data source.

3. optimal data source

Optimal data source is that background color and hair color color range are obvious respectively, and cross section is few.These all require the clear of camera, the processing of background color various factorss such as (environmental interference).The problem that hair intersects can solve, and these can carry out analyzing and processing by the valid data scope.

Data model be the key of whole software design, the analytical approach of using comprises gradation of image, corrosion, methods such as filtering, these methods all are to get rid of the effective ways that ambient color disturbs, as for diameter of hair, radical is added up these and is related to the gathering statistics, geometry, statistics, probability statistics analyses such as (confidence levels).

The end value that has obtained the radical that diameter that the otter rabbit hair sends out and the otter rabbit hair send out so far, but these numerical value still are incredible, because such data result does not also carry out data check, do not carry out the data post-processed.Because algorithm is dead, signal (image of collection) is Protean, and the result who obtains also has true value, falsity.

So-called true value is within the valid data scope, and for example the diameter of hair of an otter rabbit can not be little of 1um, also can not be greatly to 100 and even um more than 200, so data also need to carry out verification.Need desirable processing module of design to get rid of falsity, stay true value.

The first step is cut apart the algorithm that obtains diameter of hair to threshold line and is improved.The data of handling all are string arrays, are exactly a string number in simple terms, 640 integers (scope is 0～255).Handle mathematical method by signal, designed the algorithm of a calculated diameter.(MAX+MIN)/2 as the value of threshold line, may make mistakes sometimes, if MAX or MIN deviation are big especially, can cause threshold line to hope that very big or minimum deflection is excessive like this, cause diameter result of calculation bigger than normal, less than normal.Therefore, the model of design is as follows:

1. calculate maximal value MAX, minimum value MIN.

2. remove MAX, the MIN value corresponding is got back to step 1, and the data that stay are being removed maximum, and minimum so circulates after 5 times, and the MAX that removes up to the last round of MAX that removes and next round differs less than 5 and gets final product with interior.

3. Sheng Xia serial data will be the last array of calculating.The mean value of computational data string, with this mean value as threshold line.

4. add the noise removal algorithm.

So-called denoise algorithm: for example, situation shown in Figure 16 appears in data.

Can see that crossing the primary position of threshold line among the figure has a little noise, if do not add judgement, will cause noise a very little diameter value partly to occur, this will directly cause result data unavailable.View data is very abundant, and affected by environment very big, so must be got rid of.

It is as follows to get rid of algorithm, threshold line V, design threshold line V1, threshold line V2, V1=V-T, V2=V+T.The size of T is according to artificial needs, and the noise range of Pai Chuing is a little bigger if desired, and T is just a little bigger, the noise range point that need get rid of, and T is with regard to point.Calculate like this when judging then: preceding 1 R1 of rising edge data value, if back 1 R2 is R1＜=V1 and R2 〉=V2, such result has just really passed through threshold line at last, otherwise R2 is made as the R1=V1 that next time judges.Preceding 1 R1 of trailing edge data value in like manner, if back 1 R2 is R1 〉=V2 and R2＜=V1, such result has just really passed through threshold line at last, otherwise R2 is made as the R1=V2 that next time judges.The index difference of the scope point that complete rising and complete decline are exactly a diameter.This difference is exactly the shared number of pixels of diameter, and number can be scaled diameter.

Obtain the diameter of hair serial data of a width of cloth figure, should have 30 groups approximately at least, the array length of these 30 groups of data is enumerated out respectively, get rid of maximum number 5 groups, 5 groups of minimum number, remaining 20 groups as data computation.

20 groups of data are exactly 20 groups of diameter values, the number of every group of array, every group of the diameter number that data computation goes out, the just radical of hair exactly.What wherein number was smaller is the part of hair adhesion on image certainly, so just can cause the hair root to reduce.Designing following algorithm gets rid of:

1. every group of data are investigated, diameter of hair through [(the actual realization of numerical value * screen width)/640] after the ratiometric conversion greater than removing 2 more than 40, generate 2 new numerical value, less than 10 directly exclude, put into the backup array, so-called backup array is exactly the falsity table, and these falsities will be put into statistics array the inside already.

2. calculate every group of data radical value, the array of identical radical is sorted out.For example 20 5 groups, 21 6 groups, 22 7 groups

3. will have the radical of array of maximum identical radicals as the basic radical M of hair, then will be than in the M big 2, little 2 with interior array, removes minimum or maximum hair value, after the M that the array radical is put into the hair radical is the same substantially, also be grouped into maximum identical radical arrays the inside.For example: 20 5 groups, 21 6 groups, 22 7 groups, 25 2 groups.That of 25 removes for 2 groups so.Every group of data of 21 are removed a minimum value and are classified as in 20 groups, and every group of data of 22 are removed a minimum value and also are classified as in 20 groups, and such 20 group number is 18 groups.So with these 18 groups of data successively from the 1-20 root, mutual superposition gets up to remove the group number again, obtains the diameter mean value of every actual hair.G[18] [20]; In fact at last the two-dimensional array that comes to this that obtains.for（I=0,I<20,I++） { for(j=0;j<18;j++) { AVG[I]＝AVG＋G[j][I]; } AVG[I]=AVG[I]/18; }

Can obtain the array AVG[18 as next 18 elements], the hair radical is 18 so, the array value correspondence be exactly diameter of hair.

This array is exactly the true value data that obtain at last, inserts tabulation and enters post analysis.

The present invention is illustrated by above-described embodiment, but should be understood that, above-described embodiment just is used for for example and illustrative purposes, but not is intended to the present invention is limited in the described scope of embodiments.It will be appreciated by persons skilled in the art that in addition the present invention is not limited to above-described embodiment, can also make more kinds of variants and modifications according to instruction of the present invention, these variants and modifications all drop in the present invention's scope required for protection.Protection scope of the present invention is defined by the appended claims and equivalent scope thereof.

Claims (9)

1. an otter rabbit be is characterized in that comprising the steps: by wool fibre fineness, Density Detection method

A, collection are by a hair view data;

B, will in computing machine, be stored by the hair image data transmission;

C, finish by computer-processing software and detect to analyze.

2. method according to claim 1 is characterized in that adopting detector directly to gather by the hair view data from live body otter rabbit on one's body.

3. method according to claim 1 is characterized in that adopting in step c the portable data treatment facility to finish the Data Detection analysis.

4. detector according to claim 2 is DinoLite hand-held USB digit microscope.

5. the otter rabbit be is characterized in that comprising by hair image data acquiring device (1), by hair image data storage and detection analytical equipment (2) by wool fibre fineness, Density Detection system.

6. system according to claim 5 is characterized in that described image data acquiring device is (1) DinoLite hand-held USB digit microscope.

7. system according to claim 5 is characterized in that described is the portable data treatment facility by hair image data storage and detection analytical equipment (2).

8. system according to claim 6 is characterized in that described microscope is mainly by CCD(3), convex lens (4) and lighting unit (5) constitute.

9. system according to claim 7 is characterized in that described portable data treatment facility arranges promising input/output interface (6) and charge port (7).

Images