CN102663726B - Method and device for material segmentation of material accumulation images - Google Patents

Abstract

The invention provides a method and a device for material segmentation of material accumulation images. The method mainly comprises carrying out filtering processing of original material accumulation images to obtain smoothed images, and after carrying out morphological operation on the smoothed images, carrying out subtraction on the smoothed images and the original images to obtain gradient images; carrying out threshold processing on a gray value of each pixel in the smoothed images to obtain binary images; filling internal noise of the binary images through a scan-line filling method, and disconnecting connected accumulation targets in the binary images to obtain distance smoothed images; carrying out geodesic opening operation on the distance smoothed images to extract central target areas of various accumulation objects, outwardly growing in parallel from the central target areas of various accumulation objects in the gradient images to the edges of the accumulation objects to obtain boundary information of various accumulation objects. According to embodiments provided in the invention, materials in an accumulation scene can be accurately and effectively segmented.

Description

Solid accumulation image is carried out to the method and apparatus of material segmentation

Technical field

The present invention relates to electronic technology field, particularly a kind of method and apparatus solid accumulation image being carried out to material segmentation.

Background technology

Material acquisition and processing is carried out to the solid accumulation image that travelling belt is carried, obtains the segmentation result of material, for real-time, quality testing and fault diagnosis are carried out, to enterprise production process important in inhibiting to the production run of enterprise.The solid accumulation that travelling belt is carried, the different scale of material is large and form is irregular, and segment boundary information is covered by shade, and light source causes uneven distribution in collection image, almost there is no background information in whole two field picture, cause only being difficult to distinguish deposit by the colouring information of image.If the surface relief of deposit is different, coarse unsmooth, the texture information of image is also insensitive to the segmentation of deposit.Single color, textural characteristics can not split complicated deposit image preferably.Therefore, because the material piled up has irregular shape and coarse surface, the edge describing material from the image of the solid accumulation collected exactly is a problem demanding prompt solution.

The method of a kind of Image Segmentation Using to solid accumulations such as ores of the prior art is: based on the image partition method of threshold value, namely adopts gray threshold that deposit and background are divided into bianry image, realizes the segmentation to deposit according to this.But deposit overlap is piled up on travelling belt, in the bianry image after Threshold segmentation, many places deposit that is adjacent or that pile up links together, and different objects can be split into an object; And the larger noise information of deposit inside is also difficult to remove, and this object can be caused to be split into several object.

Summary of the invention

In view of the demand of application, embodiments provide a kind of method and apparatus solid accumulation image being carried out to material segmentation, to realize carrying out accurate and effective Ground Split to deposit target.

The present invention realizes by the following method:

Solid accumulation image is carried out to a method for material segmentation, comprising:

Filtering process is carried out to the original image of solid accumulation and obtains smoothed image, after described smoothed image is carried out morphological operation, subtract each other with original image and obtain gradient image;

Thresholding process is carried out to the gray-scale value of each pixel in described smoothed image and obtains bianry image;

Filled by the internal noise of Scanning-line Filling method to described bianry image, and disconnect the deposit target connected in described bianry image obtain distance smoothed image;

Geodetic opening operation is carried out to described distance smoothed image, extract the focus target region of each deposit, focus target region from each deposit on described gradient image walks abreast to outgrowth, grows into the edge of deposit always, obtains the boundary information of each deposit.

Solid accumulation image is carried out to a device for material segmentation, comprising:

Level and smooth and gradient image acquisition module, obtaining smoothed image for carrying out filtering process to the original image of solid accumulation, after described smoothed image is carried out morphological operation, subtracting each other obtain gradient image with original image;

Bianry image acquisition module, obtains bianry image for carrying out thresholding process to the gray-scale value of each pixel in described smoothed image;

Distance smoothed image acquisition module, for being filled by the internal noise of Scanning-line Filling method to described bianry image, and is disconnected the deposit target connected in described bianry image and obtains distance smoothed image;

Deposit boundary information acquisition module, for carrying out geodetic opening operation to described distance smoothed image, extract the focus target region of each deposit, focus target region from each deposit on described gradient image walks abreast to outgrowth, grow into the edge of deposit always, obtain the boundary information of each deposit.

The embodiment of the present invention obtains bianry image by carrying out thresholding process to smoothed image, the internal noise of bianry image is filled, smoothed image of adjusting the distance again carries out geodetic opening operation, extract the focus target region of each deposit, and obtain deposit border accurately, accurate and effective Ground Split can be carried out to the material piled up in scene.

Accompanying drawing explanation

In order to be illustrated more clearly in the technical scheme of the embodiment of the present invention, below the accompanying drawing used required in describing embodiment is briefly described.

A kind of concrete processing flow chart solid accumulation image being carried out to the method for material segmentation that Fig. 1 provides for the embodiment of the present invention one;

What Fig. 2 provided for the embodiment of the present invention one a kind ofly removes the noise of deposit border inner and the schematic diagram of noise region;

A kind of structural representation solid accumulation image being carried out to the device of material segmentation that Fig. 3 provides for the embodiment of the present invention one.

Embodiment

For ease of the understanding of the present invention, be further explained explanation below in conjunction with accompanying drawing for concrete embodiment, and embodiment does not form the restriction to the embodiment of the present invention.

Embodiment one

A kind of concrete treatment scheme of solid accumulation image being carried out to the method for material segmentation that this embodiment provides as shown in Figure 1, comprises following treatment step:

Step 11, filtering process is carried out to the original image of the colour of solid accumulation obtain smoothed image.

Video camera and light source are arranged on above deposit travelling belt, by kilomega network, collection image transfer are sent on remote server, carry out the real-time process of image.Every two field picture is all handled as follows:

The original image of the colour of gathered solid accumulation is carried out bilateral filtering process and obtains smoothed image, color value after each pixel filter, obtained by neighborhood territory pixel weighted sum, weight is made up of two factors, i.e. the color difference of neighborhood territory pixel and space length.This bilateral filtering process, while removing deposit internal noise, retains the boundary information of deposit substantially.

Step 12, optimum thresholding process is carried out to above-mentioned smoothed image obtain bianry image.

The smoothed image that obtains after bilateral filtering process is adopted and optimizes method that global threshold and adaptive local threshold method combine and carry out binary conversion treatment and obtain bianry image.Above-mentioned binary conversion treatment process mainly comprises:

Large and small two square window are made in the picture, the window threshold value (t of these two square window centered by each pixel p (x, y) in the smoothed image of deposit _bigand t _small) determined by maximum entropy method, select according to the principle of entropy maximization, namely that gray-scale value makes the entropy of window maximum, then select this gray-scale value as the threshold value of window.Above-mentioned t _bigbe the window threshold value of square window larger in two square window, above-mentioned t _smallit is the window threshold value of square window less in two square window.

If the yardstick of above-mentioned two square window is N*N, gray shade scale is from l ₁to l ₂, threshold value t (l ₁≤ t≤l ₂) as window threshold value time, the entropy of window is:

$H\left(t\right)=H_A\left(t\right)+H_B\left(t\right)=-\underset{i=l1}{\overset{t}{\mathrm{\Σ}}}{p}_{i,A}\left(t\right)\log p_{i,A}\left(t\right)-\underset{i=t+1}{\overset{l2}{\mathrm{\Σ}}}{p}_{i,B}\left(t\right)\log p_{i,B}\left(t\right)$ Formula 1

From l ₁to l ₂in gray level, ask the threshold value t of the entropy maximization of two windows _bigand t _small

$\left\{\begin{array}{c}{t}_{\mathrm{big}}=\arg \underset{l1\&le;t<l2}{\max }\left\{H_{\mathrm{big}}\left(t\right)\right\}\\ t_{\mathrm{small}}=\arg \underset{l1\&le;t<l2}{\max }\left\{H_{\mathrm{small}}\left(t\right)\right\}\end{array}\right.$ Formula 2

Above-mentioned H _athe entropy of the subsystem that t all pixels that () is less than or equal to t for gray-scale value are formed, H _bt () is greater than the entropy of the formation subsystem of all pixels of t for gray-scale value, p _{i, A}t () is be less than or equal in the subsystem of all pixels formations of t at gray-scale value, the distribution probability of the pixel of gray scale i, p _{i, B}t () is be greater than in the subsystem of all pixels formations of t at gray-scale value, the distribution probability of the pixel of gray scale i, H _bigt maximal value that () is H (t), H _smallt minimum value that () is H (t).

When the gray-scale value of pixel p (x, y) is more than or equal to t _bigand t _smallduring minimum value in two threshold values, be then 255 by the binarization of gray value of this pixel p (x, y), as the object pixel of deposit;

When the gray-scale value of pixel p (x, y) is less than t _bigand t _smallduring minimum value in two threshold values, be then 0 by the binarization of gray value of this pixel p (x, y), as the background pixel of deposit.

The yardstick of above-mentioned large and small two square window is determined by piling up impersonal language, and the yardstick as large window is determined according to large object, and the yardstick of wicket is determined according to wisp.

Step 13, by carrying out filling the noise information removed in above-mentioned bianry image to deposit border inner.

Described bianry image after scanning binaryzation, for each object pixel in the above-mentioned bianry image scanned, rebuild a new region and add in the chained list of region, Freeman (eight chain code methods) method is utilized to extract the closed edge of above-mentioned new region, and determine that edge pixel is left pixel, or right pixel.Then, chain code technology is adopted to be filled by the inside of Scanning-line Filling method to new region, such as, fill from left pixel to right pixel, and add up the number of filler pixels and the area of filling material, to remove the internal noise of described bianry image, namely remove the noise of deposit inside, and by design area threshold value, remove the noise region that area is little.Pixel after filling is not re-used as object pixel.

This embodiment provide a kind of remove deposit border inner noise and noise region schematic diagram as shown in Figure 2, in Fig. 2, the image on the left side is the image of the deposit before internal noise is filled, and the right is the image of the deposit after filling internal noise.

Step 14, the deposit target disconnected.

This treatment step carries out in two steps, the first step: carry out corrosion treatment to the image after filling internal noise, Erodent Algorithm is the square template of 3*3, and continuous several times (such as 3 times) corrode, the deposit of the small scale that breaking part links together; Second step: bianry image again gray processing is obtained distance smoothed image by range conversion, using the object pixel in bianry image to the bee-line of its object boundary as the gray-scale value of above-mentioned object pixel, the join domain of the deposit of large scale can be disconnected further by range conversion.

Step 15, to described distance smoothed image carry out geodetic opening operation, extract unique focus target region of each deposit.

Carry out geodetic opening operation to described distance smoothed image, extract unique focus target region of each deposit, the processing procedure of above-mentioned geodetic opening operation comprises:

1) n time is carried out to described distance smoothed image and measures corrosion treatment,

$f_D(x,y)=R_D^n\left(\min (f(x,y),B)\right)$ Formula 3

Wherein, f (x, y) is a pixel in described distance smoothed image, and B is the template subimage centered by pixel f (x, y), f _d(x, y) is the pixel f (x, y) measuring corrosion treatment and obtain for the D time, f _d(x, y) is the pixel with minimum gradation value in target B.

When measuring corrosion treatment D+1 time, with f _dtemplate subimage B is reset, f centered by (x, y) _d+1(x, y) is the pixel with minimum gradation value in the template subimage B that resets.The rest may be inferred, until obtain f to the corrosion treatment that measures for N time that smoothed image sets _n(x, y), f _n(x, y) is the pixel f (x, y) measuring corrosion treatment and obtain for the N time.Above-mentioned N, generally between 10 to 15, is determined by the color difference of object and background in smoothed image, and proportional.

2) grown by the image after mensuration corrosion treatment

① $f_E^0(x,y)=f_N^0(x,y)$

Above-mentioned the gray-scale value of pixel (x, y) during for starting to grow, for f _nthe gray-scale value of (x, y).

During the i-th secondary growth, the gray-scale value of pixel (x, y) get the gray-scale value of the maximum neighborhood of gray-scale value in eight neighborhoods of the pixel after secondary growth.

$f_E^i(x,y)=\max \left(M_8\left(f_E^{i-1}(x,y)\right)\right)$

The gray-scale value of the pixel (x, y) 2. after the i-th secondary growth and former grey scale pixel value f ⁰(x, y) compares, and gets minimum value as the i-th secondary growth result

$f_E^i(x,y)=\min (f_E^i(x,y),f^0(x,y))$

③

When time, illustrate that above-mentioned pixel (x, y) is after burn into growth process, its gray-scale value adds.So, with based on, re-start the process of above-mentioned i secondary growth, until $f_E^i(x,y)=f_E^i(x,y).$

The mensuration burn into that all pixels in above-mentioned distance smoothed image all carry out in above-mentioned geodetic opening operation is grown out, obtains reconstructed image.

3) regional area with maximum gradation value in reconstructed image is extracted, the central area of deposit in this regional area and image.

Step 16, morphological operation is carried out to the smoothed image after bilateral filtering, subtract each other with original image, obtain the gradient image of smoothed image.

Step 17, boundary marker is carried out to the focus target region of each deposit, gradient image walks abreast to outgrowth from each mark center target area, unanimously grows into the edge of deposit or run into the Edge-stopping of other deposit.

By the image after the above-mentioned geodetic opening operation of chain code technology process, the border in the focus target region of each deposit of acquisition, row bound of going forward side by side marks.By boundary information outside parallel growth on gradient image in the focus target region of the deposit after mark, unanimously grow into the edge of deposit or run into the Edge-stopping of other deposit.Namely the border of extracting growth obtains the border of each deposit.

Embodiment two

A kind of device solid accumulation image being carried out to material segmentation that this embodiment provides, its concrete structure as shown in Figure 3, comprises following module:

Level and smooth and gradient image acquisition module 31, obtaining smoothed image for carrying out filtering process to the original image of solid accumulation, after described smoothed image is carried out morphological operation, subtracting each other obtain gradient image with original image;

Bianry image acquisition module 32, obtains bianry image for carrying out thresholding process to the gray-scale value of each pixel in described smoothed image;

Distance smoothed image acquisition module 33, for being filled by the internal noise of Scanning-line Filling method to described bianry image, and is disconnected the deposit target connected in described bianry image and obtains distance smoothed image;

Deposit boundary information acquisition module 34, for carrying out geodetic opening operation to described distance smoothed image, extract the focus target region of each deposit, focus target region from each deposit on described gradient image walks abreast to outgrowth, grow into the edge of deposit always, obtain the boundary information of each deposit.

Concrete, described bianry image acquisition module 32, also for making large and small two square window in the picture centered by each pixel p (x, y) in described smoothed image, t _bigfor the gray threshold of large scale square window, t _smallfor the gray threshold of small scale square window, described t _bigand t _smalldetermined by maximum entropy method, select according to the principle of entropy maximization;

When the gray-scale value of pixel p (x, y) is more than or equal to t _bigand t _smallduring minimum value in two threshold values, be then 255 by the binarization of gray value of this pixel p (x, y), as the object pixel of deposit;

When the gray-scale value of pixel p (x, y) is less than t _bigand t _smallduring minimum value in two threshold values, be then 0 by the binarization of gray value of this pixel p (x, y), as the background pixel of deposit.

Concrete, described distance smoothed image acquisition module 33, also for each object pixel in described bianry image, rebuild a new region, eight chain code methods are utilized to extract the edge of described new region, chain code technology is utilized to be filled by the inside of Scanning-line Filling method to described new region, to remove the internal noise of described bianry image.

Concrete, described deposit boundary information acquisition module 34, also measures corrosion treatment for carrying out N time to described distance smoothed image,

$f_D(x,y)=R_D^n\left(\min (f(x,y),B)\right)$

Wherein, f (x, y) is a pixel in described distance smoothed image, and B is the template subimage centered by pixel f (x, y), f _d(x, y) is the pixel f (x, y) measuring corrosion treatment and obtain for the D time, f _d(x, y) is the pixel with minimum gradation value in target B;

When measuring corrosion treatment D+1 time, with f _dtemplate subimage B is reset, f centered by (x, y) _d+1(x, y) is the pixel with minimum gradation value in the template subimage B that resets, and the rest may be inferred, until obtain f to N the mensuration corrosion treatment that smoothed image sets _n(x, y), f _n(x, y) is the pixel f (x, y) measuring corrosion treatment and obtain for the N time;

Grown by the image after measuring corrosion treatment

$f_E^0(x,y)=f_N^0(x,y)$

Described the gray-scale value of pixel (x, y) during for starting to grow, for f _nthe gray-scale value of (x, y);

During the i-th secondary growth, the gray-scale value of pixel (x, y) get the gray-scale value of the maximum neighborhood of gray-scale value in eight neighborhoods of the pixel after secondary growth;

$f_E^i(x,y)=\max \left(M_8\left(f_E^{i-1}(x,y)\right)\right)$

The gray-scale value of the pixel (x, y) after the i-th secondary growth and former grey scale pixel value f ⁰(x, y) compares, and gets minimum value as the i-th secondary growth result

$f_E^i(x,y)=\min (f_E^i(x,y),f^0(x,y))$

When time, with based on, re-start described growth process, until $f_E^i(x,y)=f_E^i(x,y);$

All pixels in described distance smoothed image are all carried out to the described mensuration burn into growth process in described geodetic opening operation, obtain reconstructed image;

Extract the regional area with maximum gradation value in reconstructed image, the central area of deposit in this regional area and image.

Concrete, described deposit boundary information acquisition module 34, also for by the image after geodetic opening operation described in chain code technology process, the border in the focus target region of each deposit of acquisition, row bound of going forward side by side marks;

By boundary information outside parallel growth on gradient image in the focus target region of the deposit after mark, unanimously grow into the edge of deposit or run into the Edge-stopping of other deposit, namely the border of extracting growth obtains the border of each deposit.

The device of the application embodiment of the present invention the concrete processing procedure of material segmentation is carried out to solid accumulation image and preceding method embodiment similar, repeat no more herein.

One of ordinary skill in the art will appreciate that all or part of flow process realized in above-described embodiment method, that the hardware that can carry out instruction relevant by computer program has come, described program can be stored in a computer read/write memory medium, this program, when performing, can comprise the flow process of the embodiment as above-mentioned each side method.Wherein, described storage medium can be magnetic disc, CD, read-only store-memory body (Read-Only Memory, ROM) or random store-memory body (Random Access Memory, RAM) etc.

In sum, because deposit image affects very large by apparatus for making a video recording quality and light source, and on-the-spot pollution level is high, and conveyer belt speed is fast, and the brightness of deposit image and sharpness difference are very large, and kind and the particle size differences of ore are larger.The embodiment of the present invention obtains bianry image by carrying out thresholding process to smoothed image, the internal noise of bianry image is filled, smoothed image of adjusting the distance again carries out geodetic opening operation, extract the focus target region of each deposit, and obtain deposit border accurately, accurate and effective Ground Split can be carried out to the material piled up in scene.

The embodiment of the present invention can carry out self-adaptive processing to the deposit image under Different Light, different cultivars, varying environment, and removes the noise in image.

The above; be only the embodiment of the present invention preferably embodiment; but the protection domain of the embodiment of the present invention is not limited thereto; anyly be familiar with those skilled in the art in the technical scope that the embodiment of the present invention discloses; the change that can expect easily or replacement, within the protection domain that all should be encompassed in the embodiment of the present invention.Therefore, the protection domain of the embodiment of the present invention should be as the criterion with the protection domain of claim.

Claims (8)

1. solid accumulation image is carried out to a method for material segmentation, it is characterized in that, comprising:

Filtering process is carried out to the original image of solid accumulation and obtains smoothed image, after described smoothed image is carried out morphological operation, subtract each other with original image and obtain gradient image;

Thresholding process is carried out to the gray-scale value of each pixel in described smoothed image and obtains bianry image; Wherein, the method adopting optimization global threshold and adaptive local threshold method to combine is carried out binary conversion treatment and is obtained bianry image;

Filled by the internal noise of Scanning-line Filling method to described bianry image, and disconnect the deposit target connected in described bianry image obtain distance smoothed image; Wherein, carry out filling the noise information removed in described bianry image to the internal noise of described bianry image by Scanning-line Filling method, it comprises: the described bianry image after scanning binaryzation, for each object pixel in the described bianry image scanned, rebuild a new region and add in the chained list of region, utilize eight chain code methods to extract the closed edge of described new region, and determine that edge pixel is left pixel, or right pixel; Adopt chain code technology to be filled by the inside of Scanning-line Filling method to new region again, to remove the internal noise of described bianry image, and by design area threshold value, remove the noise region that area is little;

Geodetic opening operation is carried out to described distance smoothed image, extract the focus target region of each deposit, focus target region from each deposit on described gradient image walks abreast to outgrowth, grows into the edge of deposit always, obtains the boundary information of each deposit.

2. method of solid accumulation image being carried out to material segmentation according to claim 1, is characterized in that, described carry out thresholding process to the gray-scale value of each pixel in described smoothed image and obtains bianry image and comprise:

Large and small two square window are made in the picture, t centered by each pixel p (x, y) in described smoothed image _bigfor the gray threshold of large scale square window, t _smallfor the gray threshold of small scale square window, described t _bigand t _smalldetermined by maximum entropy method, select according to the principle of the entropy maximization making window;

When the gray-scale value of pixel p (x, y) is more than or equal to t _bigand t _smallduring the minimum value of two threshold values, be then 255 by the binarization of gray value of this pixel p (x, y), as the object pixel of deposit;

When the gray-scale value of pixel p (x, y) is less than t _bigand t _smallduring minimum value in two threshold values, be then 0 by the binarization of gray value of this pixel p (x, y), as the background pixel of deposit.

3. method of solid accumulation image being carried out to material segmentation according to claim 1, is characterized in that, described carries out geodetic opening operation to described distance smoothed image, extracts the focus target region of each deposit, comprising:

1, N time is carried out to described distance smoothed image and measures corrosion treatment,

$f_D(x,y)=R_D^n\left(\min (f(x,y),B)\right)$

Wherein, f (x, y) is a pixel in described distance smoothed image, and B is the template subimage centered by pixel f (x, y), f _d(x, y) is the pixel f (x, y) measuring corrosion treatment and obtain for the D time, f _d(x, y) is the pixel with minimum gradation value in target B;

When measuring corrosion treatment D+1 time, with f _dtemplate subimage B is reset, f centered by (x, y) _d+1(x, y) is the pixel with minimum gradation value in the template subimage B that resets, and the rest may be inferred, until obtain f to N the mensuration corrosion treatment that smoothed image sets _n(x, y), f _n(x, y) is the pixel f (x, y) measuring corrosion treatment and obtain for the N time;

2, grown by the image after mensuration corrosion treatment

$f_E^0(x,y)=f_N^0(x,y)$

Described the gray-scale value of pixel (x, y) during for starting to grow, for f _nthe gray-scale value of (x, y);

During the i-th secondary growth, the gray-scale value of pixel (x, y) get the gray-scale value of the maximum neighborhood of gray-scale value in eight neighborhoods of the pixel after secondary growth;

$f_E^i(x,y)=\max \left(M_8\left(f_E^{i-1}(x,y)\right)\right)$

The gray-scale value of the pixel (x, y) after the i-th secondary growth and former grey scale pixel value f ⁰(x, y) compares, and gets minimum value as the i-th secondary growth result

$f_E^i(x,y)=\min (f_E^i(x,y),f^0(x,y))$

When time, with based on, re-start described growth process, until $f_E^i(x,y)=f_E^i(x,y);$

All pixels in described distance smoothed image are all carried out to the described mensuration burn into growth process in described geodetic opening operation, obtain reconstructed image;

Extract the regional area with maximum gradation value in reconstructed image, the central area of deposit in this regional area and image.

4. the method for solid accumulation image being carried out to material segmentation according to any one of claims 1 to 3, it is characterized in that, the described focus target region from each deposit on described gradient image walks abreast to outgrowth, grow into the edge of deposit always, obtain the boundary information of each deposit, comprising:

By the image after geodetic opening operation described in chain code technology process, the border in the focus target region of each deposit of acquisition, row bound of going forward side by side marks;

By boundary information outside parallel growth on gradient image in the focus target region of the deposit after mark, unanimously grow into the edge of deposit or run into the Edge-stopping of other deposit, namely the border of extracting growth obtains the border of each deposit.

5. solid accumulation image is carried out to a device for material segmentation, it is characterized in that, comprising:

Level and smooth and gradient image acquisition module, obtaining smoothed image for carrying out filtering process to the original image of solid accumulation, after described smoothed image is carried out morphological operation, subtracting each other obtain gradient image with original image;

Bianry image acquisition module, obtains bianry image for carrying out thresholding process to the gray-scale value of each pixel in described smoothed image; Wherein, the method adopting optimization global threshold and adaptive local threshold method to combine is carried out binary conversion treatment and is obtained bianry image;

Distance smoothed image acquisition module, for being filled by the internal noise of Scanning-line Filling method to described bianry image, and is disconnected the deposit target connected in described bianry image and obtains distance smoothed image; Wherein, carry out filling the noise information removed in described bianry image to the internal noise of described bianry image by Scanning-line Filling method, it comprises: the described bianry image after scanning binaryzation, for each object pixel in the described bianry image scanned, rebuild a new region and add in the chained list of region, utilize eight chain code methods to extract the closed edge of described new region, and determine that edge pixel is left pixel, or right pixel; Adopt chain code technology to be filled by the inside of Scanning-line Filling method to new region again, to remove the internal noise of described bianry image, and by design area threshold value, remove the noise region that area is little;

Deposit boundary information acquisition module, for carrying out geodetic opening operation to described distance smoothed image, extract the focus target region of each deposit, focus target region from each deposit on described gradient image walks abreast to outgrowth, grow into the edge of deposit always, obtain the boundary information of each deposit.

6. device solid accumulation image being carried out to material segmentation according to claim 5, is characterized in that:

Described bianry image acquisition module, also for making large and small two square window in the picture centered by each pixel p (x, y) in described smoothed image, t _bigfor the gray threshold of large scale square window, t _smallfor the gray threshold of small scale square window, described t _bigand t _smalldetermined by maximum entropy method, select according to the principle of entropy maximization;

When the gray-scale value of pixel p (x, y) is more than or equal to t _bigand t _smallduring minimum value in two threshold values, be then 255 by the binarization of gray value of this pixel p (x, y), as the object pixel of deposit;

When the gray-scale value of pixel p (x, y) is less than t _bigand t _smallduring minimum value in two threshold values, be then 0 by the binarization of gray value of this pixel p (x, y), as the background pixel of deposit.

7. device solid accumulation image being carried out to material segmentation according to claim 5, is characterized in that:

Described deposit boundary information acquisition module, also measures corrosion treatment for carrying out N time to described distance smoothed image,

$f_D(x,y)=R_D^n\left(\min (f(x,y),B)\right)$

Wherein, f (x, y) is a pixel in described distance smoothed image, and B is the template subimage centered by pixel f (x, y), f _d(x, y) is the pixel f (x, y) measuring corrosion treatment and obtain for the D time, f _d(x, y) is the pixel with minimum gradation value in target B;

When measuring corrosion treatment D+1 time, with f _dtemplate subimage B is reset, f centered by (x, y) _d+1(x, y) is the pixel with minimum gradation value in the template subimage B that resets, and the rest may be inferred, until obtain f to N the mensuration corrosion treatment that smoothed image sets _n(x, y), f _n(x, y) is the pixel f (x, y) measuring corrosion treatment and obtain for the N time;

Grown by the image after measuring corrosion treatment

$f_E^0(x,y)=f_N^0(x,y)$

Described the gray-scale value of pixel (x, y) during for starting to grow, for f _nthe gray-scale value of (x, y);

During the i-th secondary growth, the gray-scale value of pixel (x, y) get the gray-scale value of the maximum neighborhood of gray-scale value in eight neighborhoods of the pixel after secondary growth;

$f_E^i(x,y)=\max \left(M_8\left(f_E^{i-1}(x,y)\right)\right)$

The gray-scale value of the pixel (x, y) after the i-th secondary growth and former grey scale pixel value f ⁰(x, y) compares, and gets minimum value as the i-th secondary growth result

$f_E^i(x,y)=\min (f_E^i(x,y),f^0(x,y))$

When time, with based on, re-start described growth process, until $f_E^i(x,y)=f_E^i(x,y);$

All pixels in described distance smoothed image are all carried out to the described mensuration burn into growth process in described geodetic opening operation, obtain reconstructed image;

Extract the regional area with maximum gradation value in reconstructed image, the central area of deposit in this regional area and image.

8. device solid accumulation image being carried out to material segmentation according to any one of claim 5 to 7, is characterized in that:

Described deposit boundary information acquisition module, also for by the image after geodetic opening operation described in chain code technology process, the border in the focus target region of each deposit of acquisition, row bound of going forward side by side marks;

By boundary information outside parallel growth on gradient image in the focus target region of the deposit after mark, unanimously grow into the edge of deposit or run into the Edge-stopping of other deposit, namely the border of extracting growth obtains the border of each deposit.