CN106097344A - A kind of image processing method detecting geometric form impurity in rubber for tire and system - Google Patents

Abstract

The invention discloses a kind of image processing method detecting geometric form impurity in rubber for tire and system, method includes: step 1, the binary image of acquisition rubber for tire；Step 2, different characteristic value according to rubber pixel and white carbon black pixel, obtain the connected domain of white carbon black pixel in binary image；Step 3, calculate the area A and width d of each connected domain；Step 4, area A and width d according to each connected domain, calculate flexibility S of each connected domain；Step 5, the flexibility of calculated each connected domain being compared with predetermined threshold value, it is to have the geometric form impurity of white carbon black pixel in binary image that flexibility exceedes the connected domain of predetermined threshold value.The beneficial effect comprise that: utilize connected domain flexibility that binary image is processed, can identify and image splits the geometric form impurity such as straight line, circle and curve in binary image, the binary image obtained more can embody the distributivity of white carbon black in rubber.

Description

A kind of image processing method detecting geometric form impurity in rubber for tire and system

Technical field

The present invention relates to technical field of image processing, particularly relate to a kind of detect the image of geometric form impurity in rubber for tire Processing method and system.

Background technology

In and sizing material image acquisition process mixing at actual rubber for tire, uncontrollable due to some artificial, system etc. The factor of system can make to exist in sizing material sample some impurity.These impurity or be embedded in inside sizing material or be attached to the table of sizing material Face.Under the microscope of band light source during imaging, these impurity can present the optical characteristics identical with white carbon black.Identification due to white carbon black It is that therefore, these impurity also can be identified as white carbon black, table in binary map based on white carbon black and rubber difference characteristic in gray scale The now white portion as white carbon black, the identification to white carbon black brings certain difficulty.

In bianry image, the half-tone information of single pixel cannot be distinguished by white carbon black and impurity equally, it is therefore desirable to carries out Zone marker, is divided into the same area by adjacent feature pixel, forms connected domain, by analyzing the geometrical property of connected domain White carbon black and impurity are made a distinction.The geometric properties of connected domain is studied, impurity is divided into following four classes: bulky grain is miscellaneous Matter, linear impurity, pore impurity (circular impurity) and shaped form impurity, for the large granular impurity size by region Can be carried out distinguishing, but for linear impurity, circular impurity and shaped form impurity these three are made a distinction from image The most extremely difficult.

Summary of the invention

The technical problem to be solved in the present invention is for being difficult in prior art be identified geometric form impurity distinguishing Defect, it is provided that a kind of image processing method detecting geometric form impurity in rubber for tire and system.

The technical solution adopted for the present invention to solve the technical problems is:

There is provided a kind of and detect the image processing method of geometric form impurity in rubber for tire, comprise the following steps:

Step 1, the binary image of acquisition rubber for tire；

Step 2, according to the different characteristic value of rubber pixel and white carbon black pixel in binary image, obtain binary picture The connected domain of the white carbon black pixel in Xiang；

The area A and width d of each connected domain of white carbon black pixel in step 3, calculating binary image；

Step 4, area A and width d according to each connected domain, calculate flexibility S of each connected domain, computing formula For:

$S=\frac{A}{d^2};$

Step 5, the flexibility of calculated each connected domain being compared with predetermined threshold value, flexibility exceedes default The connected domain of threshold value is to have the geometric form impurity of white carbon black pixel in binary image.

In method of the present invention, described method also includes:

In step 6, image segmentation binary image, the flexibility of white carbon black pixel exceedes the connected domain of predetermined threshold value.

In method of the present invention, described predetermined threshold value is 1.3.

In method of the present invention, described step 2 specifically utilizes region-growing method to obtain white carbon black picture in binary image The connected domain of vegetarian refreshments, specifically includes following steps:

Step 21, binary image is carried out sequential scan, find that first labeled and eigenvalue is the pixel of 1 (x₀,y₀), wherein the eigenvalue of rubber pixel is 0, and the eigenvalue of white carbon black pixel is 1；

Step 22, will be with (x₀,y₀Labeled and pixel labelling that eigenvalue is 1 in 8 neighborhood territory pixel points centered by) For same connected domain and be pressed in stack；

Step 23, when determining that stack, for time empty, takes out a pixel (x from stack_a,y_a), by pixel (x_a,y_a) make For (x₀,y₀) and go to step 22；

Step 24, when determining that stack is sky, it is judged that the pixel that all eigenvalues the are 1 the most all quilts in binary image Labelling, if it is not, then go to step 21；The most then complete the labelling of connected domain.

In method of the present invention, in described step 3, the computing formula of the area A calculating each connected domain is:

$A=\underset{i=1}{\overset{M}{\Σ}}\underset{j=1}{\overset{N}{\Σ}}f(x,y),$

A is the area of connected domain, and (x, y) is pixel to f, and M is the scope in connected domain x direction, and N is connected domain y direction Scope.

In method of the present invention, described step 3 specifically utilize erosion algorithm calculate the width d of each connected domain, Specifically include following steps:

Step 31, utilizing structural elements G to be scanned each connected domain, wherein G is the structural elements of 3*3 size, the center of G Point is (x_b,y_b)；

Step 32, judge in each connected domain that structural elements G covers, whether the eigenvalue of all pixels is 1, wherein The eigenvalue of rubber pixel is 0, and the eigenvalue of white carbon black pixel is 1:

The most then keep pixel (x_b,y_b) eigenvalue be 1；

If it is not, then by pixel (x_b,y_b) eigenvalue be assigned to 0；

Step 33, when determining the single pass being not fully complete each connected domain, go to step 31；

Step 34, when having determined the single pass to each connected domain, corrosion number of times p be assigned to p+1, it is judged that Mei Gelian In logical territory, whether the eigenvalue of all pixels is 0, and wherein the initial value of corrosion number of times p is 0；

Step 35, when determining that in each connected domain, the eigenvalue of all pixels is not 0, go to step 31；

Step 36, when determining that in each connected domain, the eigenvalue of all pixels is 0, according to corrosion number of times p calculate The width d of each connected domain, computing formula is d=2p.

The present invention also provides for a kind of detecting the image processing system of geometric form impurity in rubber for tire, specifically includes:

First acquisition module, for obtaining the binary image of rubber for tire；

Second acquisition module, is used for according to the different characteristic value of rubber pixel and white carbon black pixel in binary image, The connected domain of the white carbon black pixel in acquisition binary image；

First computing module, for calculating area A and the width of each connected domain of white carbon black pixel in binary image d；

Second computing module, for area A and width d according to each connected domain, calculates the flexibility of each connected domain S, computing formula is:

$S=\frac{A}{d^2};$

Comparison module, for the flexibility of calculated each connected domain is compared with predetermined threshold value, flexibility The connected domain exceeding predetermined threshold value is to have the geometric form impurity of white carbon black pixel in binary image.

In system of the present invention, described system also includes:

Segmentation module, in image segmentation binary image, the flexibility of white carbon black pixel exceedes the connection of predetermined threshold value Territory.

In system of the present invention, described predetermined threshold value is 1.3.

In system of the present invention, described second acquisition module includes:

First scanning element, for binary image is carried out sequential scan, finds first not labeled and eigenvalue It is the pixel (x of 1₀,y₀), wherein the eigenvalue of rubber pixel is 0, and the eigenvalue of white carbon black pixel is 1；

Indexing unit, being used for will be with (x₀,y₀8 neighborhood territory pixel points centered by) are not labeled and eigenvalue is the pixel of 1 Point is labeled as same connected domain and is pressed in stack；

First determines unit, for when determining that stack, not for time empty, takes out a pixel (x from stack_a,y_a), by pixel (x_a,y_a) as (x₀,y₀) and return label unit；

Second determines unit, for when determining that stack is sky, it is judged that the pixel that all eigenvalues are 1 in binary image Point is the most labeled, if it is not, then return the first scanning element；The most then complete the labelling of connected domain.

In system of the present invention, the computing formula of the area A that described first computing module calculates each connected domain is:

$A=\underset{i=1}{\overset{M}{\Σ}}\underset{j=1}{\overset{N}{\Σ}}f(x,y),$

A is the area of connected domain, and (x, y) is pixel to f, and M is the scope in connected domain x direction, and N is connected domain y direction Scope.

In system of the present invention, described first computing module includes:

Second scanning element, is used for utilizing structural elements G to be scanned each connected domain, and wherein G is the structure of 3*3 size Unit, the central point of G is (x_b,y_b)；

Judging unit, for judging in each connected domain that structural elements G covers, whether the eigenvalue of all pixels is 1, wherein the eigenvalue of rubber pixel is 0, and the eigenvalue of white carbon black pixel is 1: the most then keep pixel (x_b,y_b) spy Value indicative is 1；If it is not, then by pixel (x_b,y_b) eigenvalue be assigned to 0；

3rd determines unit, for when determining the single pass being not fully complete each connected domain, returns the second scanning single Unit；

4th determines unit, and for when having determined the single pass to each connected domain, corrosion number of times p is assigned to p+1, Judging in each connected domain, whether the eigenvalue of all pixels is 0, wherein the initial value of corrosion number of times p is 0；

5th determines unit, for when determining that in each connected domain, the eigenvalue of all pixels is not 0, returns the Two scanning elements；

6th determines unit, for when determining that in each connected domain, the eigenvalue of all pixels is 0, according to corrosion Number of times p calculates the width d of each connected domain, and computing formula is d=2p.

The beneficial effect comprise that: utilize connected domain flexibility that binary image is processed, can identify And the geometric form impurity such as straight line, circle and curve in image segmentation binary image, the binary image obtained more can embody The distributivity of white carbon black in rubber.

Accompanying drawing explanation

Below in conjunction with drawings and Examples, the invention will be further described, in accompanying drawing:

Fig. 1 is that the embodiment of the present invention a kind of detects the flow process of the image processing method of geometric form impurity in rubber for tire and show It is intended to；

Fig. 2 is that the white carbon black pixel in binary image is connected by the region-growing method that utilizes in the embodiment of the present invention The schematic flow sheet of the method for the labelling in territory；

Fig. 3 is the flow process signal of the method for the width utilizing the erosion algorithm each connected domain of calculating in the embodiment of the present invention Figure；

Fig. 4 is that a kind of of the present invention detects the structural representation of the image processing system of geometric form impurity in rubber for tire；

Fig. 5 (a) is large granular impurity image；

Fig. 5 (b) is linear impurity image；

Fig. 5 (c) is pore impurity image；

Fig. 5 (d) is curve impurity image；

Fig. 6 (a) is the large granular impurity binary image that Fig. 5 (a) is corresponding；

Fig. 6 (b) is the linear impurity binary image that Fig. 5 (b) is corresponding；

Fig. 6 (c) is the pore impurity binary image that Fig. 5 (c) is corresponding；

Fig. 6 (d) is the curve impurity binary image that Fig. 5 (d) is corresponding；

Fig. 7 is the region stretching schematic diagram in the embodiment of the present invention；

Fig. 8 is the corrosion schematic diagram in the embodiment of the present invention；

Fig. 9 is the flexibility statistical result chart in the embodiment of the present invention；

Figure 10 is the experimental result comparison diagram in the embodiment of the present invention.

Detailed description of the invention

In order to make the purpose of the present invention, technical scheme and advantage clearer, below in conjunction with drawings and Examples, right The present invention is further elaborated.Should be appreciated that specific embodiment described herein only in order to explain the present invention, not For limiting the present invention.

The embodiment of the present invention is a kind of detects the image processing method of geometric form impurity in rubber for tire, as it is shown in figure 1, bag Include:

Step 1, the binary image of acquisition rubber for tire；

Step 2, according to the different characteristic value of rubber pixel and white carbon black pixel in binary image, obtain binary picture The connected domain of the white carbon black pixel in Xiang；

The area A and width d of each connected domain of white carbon black pixel in step 3, calculating binary image；

Step 4, area A and width d according to each connected domain, calculate flexibility S of each connected domain, computing formula For:

$S=\frac{A}{d^2};$

Step 5, the flexibility of calculated each connected domain being compared with predetermined threshold value, flexibility exceedes default The connected domain of threshold value is to have the geometric form impurity of white carbon black pixel in binary image.

As shown in Fig. 5 (a)-(d), the geometric properties of connected domain is studied by the present invention, and impurity is divided into following four classes: Large granular impurity, linear impurity, pore impurity (circular impurity) and shaped form impurity, the binary picture that these four impurity is corresponding As shown in Fig. 6 (a)-(d).By study of tire rubber image cathetus shape impurity, shaped form impurity and circular impurity, find These three impurity has the length of curve of a common feature, i.e. connected domain and different widths very big, and this makes connected domain exist Geometrically present the feature of " elongated ".Therefore, the present invention proposes the concept of a kind of flexibility and enters the geometric form impurity in rubber Row is distinguished and identifies.

The definition of connected domain flexibility:

$S=\frac{l}{d}$

Wherein, l is the length of curve of connected domain, is not the straight length of connected domain head and the tail, and d is the width of connected domain.By Length of curve in connected domain cannot directly calculate, and therefore stretches region, as shown in Figure 6, at region area and width Under conditions of constant, region is drawn into rectangle, ignores the difference between width, then the computational methods of length of curve l are as follows:

$l=\frac{A}{d}$

Wherein, A is the area of connected domain, the number of the pixel comprised in can being expressed as connected domain.Expression formula is entered Row integrates the further expression formula that i.e. can get flexibility, as follows:

$S=\frac{A}{d^2}$

In the embodiment of the present invention, also include:

In step 6, image segmentation binary image, the flexibility of white carbon black pixel exceedes the connected domain of predetermined threshold value, is i.e. It is partitioned into the geometric form impurity of white carbon black pixel in binary image.

In the embodiment of the present invention, predetermined threshold value is 1.3.

As shown in Figure 8, the size of the predetermined threshold value of connected domain flexibility in enforcement one is determined by the method for experiment.Choose Ten grade standard white carbon black images ten, every one-level selects one, owing to the length of shaped form impurity is bigger, in order to simplify calculating choosing Select front 100 connected domains that in binary map, region curve length is maximum and carry out contrast experiment, flexibility is divided into 10 grades, The number recording the connected domain comprised in each grade obtains Fig. 8.

Analysis chart 8 understands, and in ten grade standard pictures, the value of flexibility is all below 1.3, and the b in Fig. 4, c, d tri-figure There is the situation more than 1.3 in the value making flexibility owing to there is linear impurity, therefore arranges presetting of connected domain flexibility Threshold value is 1.3.

In the embodiment of the present invention, use region-growing method find and labelling binary image in connected domain, such as Fig. 2 institute Showing, step 2 specifically includes:

Step 21, binary image is carried out sequential scan, find that first labeled and eigenvalue is the pixel of 1 (x₀,y₀), wherein the eigenvalue of rubber pixel is 0, and the eigenvalue of white carbon black pixel is 1；

Step 22, will be with (x₀,y₀Labeled and pixel labelling that eigenvalue is 1 in 8 neighborhood territory pixel points centered by) For same connected domain and be pressed in stack；

Step 23, when determining that stack, for time empty, takes out a pixel (x from stack_a,y_a), by pixel (x_a,y_a) make For (x₀,y₀) and go to step 22；

Step 24, when determining that stack is sky, it is judged that the pixel that all eigenvalues the are 1 the most all quilts in binary image Labelling, if it is not, then go to step 21；The most then complete the labelling of connected domain.

In method of the present invention, in described step 3, the computing formula of the area A calculating each connected domain is:

$A=\underset{i=1}{\overset{M}{\Σ}}\underset{j=1}{\overset{N}{\Σ}}f(x,y),$

A is the area of connected domain, and (x, y) is pixel to f, and M is the scope in connected domain x direction, and N is connected domain y direction Scope.

In the embodiment of the present invention, use region-growing method find and labelling binary image in connected domain, such as Fig. 2 institute Showing, step 2 specifically includes:

Step 21, carries out sequential scan to binary image, finds first not to be labeled and eigenvalue is the pixel of 1 (x₀,y₀)；

Step 22, will be with (x₀,y₀Labeled and pixel labelling that eigenvalue is 1 in 8 neighborhood territory pixel points centered by) For same connected domain and be pressed in stack；

Step 23, when determining that stack, for time empty, takes out a pixel (x from stack_a,y_a), by pixel (x_a,y_a) make For (x₀,y₀) and go to step 22；

Step 24, when determining that stack is sky, it is judged that the pixel that all eigenvalues the are 1 the most all quilts in binary image Labelling, if it is not, then go to step 21；The most then complete the labelling of connected domain.

Region-growing method, the successively each pixel in scanning binary image, when finding certain unlabelled target picture During element, it is pressed into storehouse and starts repeatedly its field of labelling from this point, until storehouse is empty.Under 8 field connectivity criteria, region The scanning times of growth method can be dropped to average 4 times by 8 times, has been truly realized quickly, simply.

In one embodiment of the present of invention, Two-Pass two-pass scan method, Seed-Filling seed filling can be passed through Method, sequential scan labelling method, backtracking passing marker method, two-way it is repeatedly scanned with method, based on binary image special representation method Labeling algorithm as based on run long representation, represent based on the distance of swimming, based on labeling methods such as Quadtrees for Representings, and be exclusively used in special The parallel labeling algorithm of the computer of architecture, find and labelling binary image in connected domain.

In the embodiment of the present invention, in step 3, the computing formula of the area A calculating each connected domain is:

$A=\underset{i=1}{\overset{M}{\Σ}}\underset{j=1}{\overset{N}{\Σ}}f(x,y),$

A is the area of connected domain, and (x, y) is pixel to f, and M is the scope in connected domain x direction, and N is connected domain y direction Scope.

The area of connected domain is obtained by calculating the number of pixel in connected domain, quick and convenient.One reality of the present invention Executing in example, the place the most different from above-described embodiment is, uses matlab function bwarea to calculate the area of connected domain.

In the embodiment of the present invention, as it is shown on figure 3, step 3 specifically utilizes erosion algorithm calculate the width of each connected domain D, specifically includes following steps:

Step 31, utilizing structural elements G to be scanned each connected domain, wherein G is the structural elements of 3*3 size, the center of G Point is (x_b,y_b)；

Step 32, judge in each connected domain that structural elements G covers, whether the eigenvalue of all pixels is 1, wherein The eigenvalue of rubber pixel is 0, and the eigenvalue of white carbon black pixel is 1:

The most then keep pixel (x_b,y_b) eigenvalue be 1；

If it is not, then by pixel (x_b,y_b) eigenvalue be assigned to 0；

Step 33, when determining the single pass being not fully complete each connected domain, go to step 31；

Step 34, when having determined the single pass to each connected domain, corrosion number of times p be assigned to p+1, it is judged that Mei Gelian In logical territory, whether the eigenvalue of all pixels is 0, and wherein the initial value of corrosion number of times p is 0；

Step 35, when determining that in each connected domain, the eigenvalue of all pixels is not 0, go to step 31；

Step 36, when determining that in each connected domain, the eigenvalue of all pixels is 0, according to corrosion number of times p calculate The width d of each connected domain, computing formula is d=2p.

As it is shown in fig. 7, A represents a set, B is a structural elements, and the expression formula that corrosion calculates is as follows:

Analysis chart 7 understands, and erosion algorithm can cut down region contour according to the size of structural elements, and not by regional morphology Impact.Therefore, region is corroded by the structural elements selecting size to be 3*3 when zoning width, often through once corroding, The width in region reduces the size of 2 pixels, until region is all corroded complete, if the step of corrosion is p, then and the width in region Degree is 2p, and the error maximum calculated is less than 2 pixel sizes, is negligible.

Following two algorithm may be applicable to calculate connected domain width: based on structural element decompose erosion algorithm, can To greatly reduce pixel access times；Erosion algorithm based on manhatton distance, can " pretreatment, repeatedly a multiplexing ", in advance The expense processed only processes generation for the first time, and all operations afterwards is all thresholding process, and thresholding then has only to The access expense of width*height.

Randomly select rubber for tire actually detected in the image of geometry impurity, its binary image utilizes region raw Regular way carries out connected component labeling, then utilizes flexibility to filter, as it is shown in figure 9, wherein (a) (c) is original image, (b) D () is the binary image being partitioned into linear impurity, the binary picture of (a) seems (b), and the binary picture of (c) seems (d).

Analysis chart 9 understands, the binary image after utilizing flexibility to process substantially be partitioned into straight line, circle and curve this three Planting geometry impurity, the binary image extracted more can embody the distributivity of white carbon black in rubber.

The embodiment of the present invention is a kind of detects the image processing system of geometric form impurity in rubber for tire, based on above-mentioned a kind of inspection The image processing method of geometric form impurity in measuring wheel tire rubber, specifically includes:

First acquisition module 1, for obtaining the binary image of rubber for tire；

Second acquisition module 2, is used for according to the different characteristic value of rubber pixel and white carbon black pixel in binary image, The connected domain of the white carbon black pixel in acquisition binary image；

First computing module 3, for calculating area A and the width of each connected domain of white carbon black pixel in binary image d；

Second computing module 4, for area A and width d according to each connected domain, calculates the flexibility of each connected domain S, computing formula is:

$S=\frac{A}{d^2};$

Comparison module 5, for the flexibility of calculated each connected domain is compared with predetermined threshold value, flexibility The connected domain exceeding predetermined threshold value is to have the geometric form impurity of white carbon black pixel in binary image.

In the embodiment of the present invention, system also includes:

Segmentation module 6, exceedes the connected domain of predetermined threshold value for splitting the flexibility of white carbon black pixel in binary image.

In the embodiment of the present invention, predetermined threshold value is 1.3.

In the embodiment of the present invention, the second acquisition module 2 includes:

First scanning element 21, for binary image is carried out sequential scan, finds first not labeled and feature Value is the pixel (x of 1₀,y₀), wherein the eigenvalue of rubber pixel is 0, and the eigenvalue of white carbon black pixel is 1；

Indexing unit 22, being used for will be with (x₀,y₀8 neighborhood territory pixel points centered by) are not labeled and eigenvalue is the picture of 1 Vegetarian refreshments is labeled as same connected domain and is pressed in stack；

First determines unit 23, for when determining that stack, not for time empty, takes out a pixel (x from stack_a,y_a), by pixel Point (x_a,y_a) as (x₀,y₀) and return label unit 22；

Second determines unit 24, for when determining that stack is sky, it is judged that the picture that all eigenvalues are 1 in binary image Vegetarian refreshments is the most labeled, if it is not, then return the first scanning element 21；The most then complete the labelling of connected domain.

In the embodiment of the present invention, the computing formula of the area A that the first computing module 3 calculates each connected domain is:

$A=\underset{i=1}{\overset{M}{\Σ}}\underset{j=1}{\overset{N}{\Σ}}f(x,y),$

A is the area of connected domain, and (x, y) is pixel to f, and M is the scope in connected domain x direction, and N is connected domain y direction Scope.

In the embodiment of the present invention, the first computing module 3 includes:

Second scanning element 31, is used for utilizing structural elements G to be scanned each connected domain, and wherein G is the knot of 3*3 size Constitutive element, the central point of G is (x_b,y_b)；

Judging unit 32, for judging that in each connected domain that structural elements G covers, the eigenvalue of all pixels is the most equal Being 1, wherein the eigenvalue of rubber pixel is 0, and the eigenvalue of white carbon black pixel is 1: the most then keep pixel (x_b,y_b) Eigenvalue is 1；If it is not, then by pixel (x_b,y_b) eigenvalue be assigned to 0；

3rd determines unit 33, for when determining the single pass being not fully complete each connected domain, returns the second scanning Unit 31；

4th determines unit 34, and for when having determined the single pass to each connected domain, corrosion number of times p is assigned to p+ 1, it is judged that in each connected domain, whether the eigenvalue of all pixels is 0, wherein the initial value of corrosion number of times p is 0；

5th determines unit 35, for when determining that in each connected domain, the eigenvalue of all pixels is not 0, returns Second scanning element 31；

6th determines unit 36, for when determining that in each connected domain, the eigenvalue of all pixels is 0, according to corruption Erosion number of times p calculates the width d of each connected domain, and computing formula is d=2p.

It should be appreciated that for those of ordinary skills, can be improved according to the above description or be converted, And all these modifications and variations all should belong to the protection domain of claims of the present invention.

Claims (12)

1. one kind is detected the image processing method of geometric form impurity in rubber for tire, it is characterised in that comprise the following steps:

Step 1, the binary image of acquisition rubber for tire；

Step 2, according to the different characteristic value of rubber pixel and white carbon black pixel in binary image, find in binary image The connected domain of white carbon black pixel；

The area A and width d of each connected domain of white carbon black pixel in step 3, calculating binary image；

Step 4, area A and width d according to each connected domain, calculate flexibility S of each connected domain, and computing formula is:

$S=\frac{A}{d^2};$

Step 5, the flexibility of calculated each connected domain being compared with predetermined threshold value, flexibility exceedes predetermined threshold value Connected domain be that binary image has the geometric form impurity of white carbon black pixel.

2. image processing method as claimed in claim 1, it is characterised in that described method also includes:

In step 6, image segmentation binary image, the flexibility of white carbon black pixel exceedes the connected domain of predetermined threshold value.

3. image processing method as claimed in claim 1, it is characterised in that described predetermined threshold value is 1.3.

4. image processing method as claimed in claim 1, it is characterised in that described step 2 specifically utilizes region-growing method to obtain Take the connected domain of white carbon black pixel in binary image, specifically include following steps:

Step 21, binary image is carried out sequential scan, find first labeled and that eigenvalue is 1 pixel (x₀, y₀), wherein the eigenvalue of rubber pixel is 0, and the eigenvalue of white carbon black pixel is 1；

Step 22, will be with (x₀,y₀In 8 neighborhood territory pixel points centered by), labeled and that eigenvalue is 1 pixel is labeled as same One connected domain is also pressed in stack；

Step 23, when determining that stack, for time empty, takes out a pixel (x from stack_a,y_a), by pixel (x_a,y_a) as (x₀, y₀) and go to step 22；

Step 24, when determining that stack is sky, it is judged that the pixel that all eigenvalues are 1 in binary image is the most all marked Note, if it is not, then go to step 21；The most then complete the labelling of connected domain.

5. image processing method as claimed in claim 1, it is characterised in that in described step 3, calculate the face of each connected domain The computing formula of long-pending A is:

$A=\underset{i=1}{\overset{M}{\Σ}}\underset{j=1}{\overset{N}{\Σ}}f(x,y),$

A is the area of connected domain, and (x, y) is pixel to f, and M is the scope in connected domain x direction, and N is the scope in connected domain y direction.

6. image processing method as claimed in claim 1, it is characterised in that specifically utilize erosion algorithm meter in described step 3 Calculate the width d of each connected domain, specifically include following steps:

Step 31, utilizing structural elements G to be scanned each connected domain, wherein G is the structural elements of 3*3 size, and the central point of G is (x_b,y_b)；

Step 32, judge in each connected domain that structural elements G covers, whether the eigenvalue of all pixels is 1, wherein rubber The eigenvalue of pixel is 0, and the eigenvalue of white carbon black pixel is 1:

The most then keep pixel (x_b,y_b) eigenvalue be 1；

If it is not, then by pixel (x_b,y_b) eigenvalue be assigned to 0；

Step 33, when determining the single pass being not fully complete each connected domain, go to step 31；

Step 34, when having determined the single pass to each connected domain, corrosion number of times p be assigned to p+1, it is judged that each connected domain In the eigenvalue of all pixels whether be 0, wherein the initial value of corrosion number of times p is 0；

Step 35, when determining that in each connected domain, the eigenvalue of all pixels is not 0, go to step 31；

Step 36, when determining that in each connected domain, the eigenvalue of all pixels is 0, according to corrosion number of times p calculate each The width d of connected domain, computing formula is d=2p.

7. one kind is detected the image processing system of geometric form impurity in rubber for tire, it is characterised in that including:

First acquisition module, for obtaining the binary image of rubber for tire；

Second acquisition module, for according to the different characteristic value of rubber pixel and white carbon black pixel in binary image, obtains The connected domain of the white carbon black pixel in binary image；

First computing module, for calculating the area A and width d of each connected domain of white carbon black pixel in binary image；

Second computing module, for area A and width d according to each connected domain, calculates flexibility S of each connected domain, meter Calculation formula is:

$S=\frac{A}{d^2};$

Comparison module, for the flexibility of calculated each connected domain being compared with predetermined threshold value, flexibility exceedes The connected domain of predetermined threshold value is to have the geometric form impurity of white carbon black pixel in binary image.

8. image processing system as claimed in claim 7, it is characterised in that described system also includes:

Segmentation module, exceedes the connected domain of predetermined threshold value for splitting the flexibility of white carbon black pixel in binary image.

9. image processing system as claimed in claim 7, it is characterised in that described predetermined threshold value is 1.3.

10. image processing system as claimed in claim 7, it is characterised in that described second acquisition module includes:

First scanning element, for binary image is carried out sequential scan, finds first not to be labeled and eigenvalue is 1 Pixel (x₀,y₀), wherein the eigenvalue of rubber pixel is 0, and the eigenvalue of white carbon black pixel is 1；

Indexing unit, being used for will be with (x₀,y₀Labeled and pixel mark that eigenvalue is 1 in 8 neighborhood territory pixel points centered by) It is designated as same connected domain and is pressed in stack；

First determines unit, for when determining that stack, not for time empty, takes out a pixel (x from stack_a,y_a), by pixel (x_a, y_a) as (x₀,y₀) and return label unit；

Second determines unit, for when determining that stack is sky, it is judged that the pixel that all eigenvalues are 1 in binary image is No the most labeled, if it is not, then return the first scanning element；The most then complete the labelling of connected domain.

11. image processing systems as claimed in claim 7, it is characterised in that described first computing module calculates each connection The computing formula of the area A in territory is:

$A=\underset{i=1}{\overset{M}{\Σ}}\underset{j=1}{\overset{N}{\Σ}}f(x,y),$

A is the area of connected domain, and (x, y) is pixel to f, and M is the scope in connected domain x direction, and N is the scope in connected domain y direction.

12. image processing systems as claimed in claim 7, it is characterised in that described first computing module includes:

Second scanning element, is used for utilizing structural elements G to be scanned each connected domain, and wherein G is the structural elements of 3*3 size, G Central point be (x_b,y_b)；

Judging unit, for judging in each connected domain that structural elements G covers, whether the eigenvalue of all pixels is 1, its The eigenvalue of middle rubber pixel is 0, and the eigenvalue of white carbon black pixel is 1: the most then keep pixel (x_b,y_b) eigenvalue It is 1；If it is not, then by pixel (x_b,y_b) eigenvalue be assigned to 0；

3rd determines unit, for when determining the single pass being not fully complete each connected domain, returns the second scanning element；

4th determines unit, and for when having determined the single pass to each connected domain, corrosion number of times p is assigned to p+1, it is judged that In each connected domain, whether the eigenvalue of all pixels is 0, and wherein the initial value of corrosion number of times p is 0；

5th determines unit, for when determining that in each connected domain, the eigenvalue of all pixels is not 0, returns second and sweeps Retouch unit；

6th determines unit, for when determining that in each connected domain, the eigenvalue of all pixels is 0, according to corrosion number of times P calculates the width d of each connected domain, and computing formula is d=2p.