A method for identifying abnormalities in injection molding of thermoplastic elastomers

Abstract

The present invention relates to the technical field of image data processing, and proposes a method for identifying abnormalities in injection molding of thermoplastic elastomers, comprising: collecting RGB images of the surface of thermoplastic elastomer injection molding parts, obtaining HSV images of the surface of injection molding parts; obtaining abnormal image blocks of injection molding parts according to the color changes in the neighborhood of pixel points in the HSV images of the surface of injection molding parts; obtaining the main extension direction of gas mark defects according to the Hessian matrix of the abnormal image blocks of injection molding parts, screening abnormal extension straight lines of injection molding parts, obtaining the gas mark regularity of the abnormal extension straight lines of injection molding parts according to the changes of pixel points on the straight lines and surrounding pixel points; judging the overall quality of injection molding parts according to the gas mark regularity. The present invention effectively avoids the disadvantage of poor accuracy in abnormal gas mark identification caused by the inability of traditional LSD algorithms to judge the extension direction of abnormal gas marks, and improves the efficiency and accuracy of abnormal gas mark detection.

Description

Technical Field

[0001] The invention relates to the technical field of image data processing, and in particular to a method for identifying anomalies in injection molding of a thermoplastic elastomer. Background Art

[0002] Thermoplastic elastomers are called "third generation rubber" after natural rubber and synthetic rubber. Thermoplastic elastomers are a kind of polymer material that has both rubber elasticity and plastic plasticity. They have many excellent properties such as simple processing and molding, aging resistance, high elasticity and relatively low production cost. Therefore, they are widely used in various industrial production fields such as automotive parts, sports equipment and aerospace. Thermoplastic elastomers are processed by injection molding to obtain parts that meet practical applications. However, during the injection molding process, due to the relatively complex injection molding process, the improper setting of the injection molding process parameters during the injection molding process leads to abnormal air marks on the surface of the injection molded parts, which has a great impact on the overall quality of the injection molded parts.

[0003] When performing abnormal gas mark defect detection on the surface images of injection molded parts, the traditional LSD algorithm cannot determine the extension direction of the abnormal gas marks in the image, and it is easy to misjudge the straight lines in the image, thereby reducing the accuracy of abnormal gas mark detection. Summary of the invention

[0004] The present invention provides a method for identifying anomalies in injection molding of thermoplastic elastomers to solve the problem that the LSD algorithm has poor recognition of the main direction of air marks. The technical solution adopted is as follows:

[0005] An embodiment of the present invention provides a method for identifying anomalies in injection molding of a thermoplastic elastomer, the method comprising the following steps:

[0006] Collect RGB images of the surface of the injection molded parts and obtain HSV images of the RGB images;

[0007] In the HSV image of the surface of the injection molded part, the color change characteristics of each pixel are obtained according to the change of the H channel value in the neighborhood of each pixel; each superpixel segmentation image block is obtained according to the color change characteristics of each pixel; the local uniform characteristic coefficient of each superpixel segmentation image block is obtained according to the difference in the color change characteristics of the pixel; the abnormal image block of the injection molded part in the superpixel segmentation image block is obtained by screening according to the local uniform characteristic coefficient;

[0008] The main extension direction of the gas mark defect in the abnormal image block of the injection molded part is obtained according to the color change characteristics of the pixel points in the abnormal image block of the injection molded part; the abnormal extension straight lines of each injection molded part in the abnormal image block of the injection molded part are obtained according to the main extension direction of the gas mark defect and the LSD straight line detection algorithm; the gas mark regularity of each abnormal extension straight line of the injection molded part is obtained according to the difference in color change characteristics between the pixel points on each abnormal extension straight line of the injection molded part and the surrounding pixel points;

[0009] The overall quality of the injection molded parts can be obtained based on the regularity of the air marks on all abnormally extended straight lines of the injection molded parts.

[0010] Preferably, obtaining the color change feature of each pixel point according to the change of the H channel value in the neighborhood of each pixel point specifically includes:

[0011] For the neighborhood of pixel point i, calculate the mean and variance of all H channel values ​​in the neighborhood; calculate the absolute value of the difference between the maximum and minimum values ​​of the H channel values ​​in the neighborhood; calculate the ratio of the variance to the mean; and use the product of the ratio and the absolute value of the difference as the color change feature of pixel point i.

[0012] Preferably, the method of obtaining each superpixel segmented image block according to the color change characteristics of each pixel point is specifically as follows: an image composed of the color change characteristics of all pixel points is used as a color change characteristic image; and the color change characteristic image is subjected to superpixel segmentation by a superpixel segmentation algorithm to obtain each superpixel segmented image block.

[0013] Preferably, obtaining the local uniform characteristic coefficient of each super-pixel segmented image block according to the color change characteristic difference of the pixel points specifically includes:

[0014] Obtain the intersection of the segmentation lines that divide the area of ​​the superpixel segmented image block into equal parts; in each superpixel segmented image block, calculate the Euclidean distance between each pixel point and the intersection point; calculate the product of the value of the Gaussian distribution function of the Euclidean distance and the color change feature of each pixel point; and use the mean value of the product of all pixels as the local uniform feature coefficient of the superpixel segmented image block.

[0015] Preferably, the abnormal image blocks of injection molded parts in the superpixel segmented image blocks are obtained by screening according to the local uniform feature coefficient, specifically:

[0016] The difference between the maximum and minimum values ​​in the color change characteristics of all super-pixel segmented image blocks is used as the judgment threshold; the difference between the local uniform feature coefficient of each super-pixel segmented image block and the color change characteristics of the central pixel point of the shape is calculated; the super-pixel segmented image block with the difference greater than or equal to the judgment threshold is used as the abnormal image block of the injection molded part.

[0017] Preferably, the main extension direction of the gas mark defect in the abnormal image block of the injection molded part is obtained according to the color change characteristics of the pixel points in the abnormal image block of the injection molded part, specifically:

[0018] A matrix composed of color change features of all pixels in the abnormal image block of each injection molded part is used as the abnormal change matrix of each injection molded part, wherein the empty values ​​in the abnormal change matrix of the injection molded parts are filled with the mean value of the color change feature values ​​of all pixels in the abnormal image block of the injection molded parts; the Hessian matrix of the abnormal change matrix of each injection molded part is obtained; the different eigenvalues ​​of each Hessian matrix are obtained by the eigenvalue decomposition algorithm; the direction of the eigenvector corresponding to the minimum eigenvalue of each Hessian matrix is ​​used as the main extension direction of the air mark defect of the abnormal image block of the injection molded part.

[0019] Preferably, the abnormal extension straight lines of each injection molded part in the abnormal image block of the injection molded part are obtained according to the main extension direction of the air mark defect and the LSD straight line detection algorithm, specifically: each straight line in each abnormal image block of the injection molded part is obtained by the LSD straight line detection algorithm; and the straight line in each abnormal image block of the injection molded part having the same direction as the main extension direction is used as the abnormal extension straight line of each injection molded part in each abnormal image block of the injection molded part.

[0020] Preferably, the gas pattern regularity of the abnormal extension straight line of each injection molded part is obtained according to the color change characteristic difference between the pixel point on the abnormal extension straight line of each injection molded part and the surrounding pixel points, specifically:

[0021] The sequence composed of the color change characteristics of all pixel points on the abnormal extension straight line of each injection molded part is taken as the central sequence, and the sequence composed of the color change characteristics of the pixel points on both sides of the straight line is recorded as the first and second regular sequences respectively; the regularity of the air marks on the abnormal extension straight line of each injection molded part is obtained according to the difference between the first and second regular sequences and the central sequence.

[0022] Preferably, the step of obtaining the regularity of the air marks of the abnormally extended straight lines of each injection molded part according to the difference between the first and second regular sequences and the central sequence specifically includes:

[0023] For each abnormally extended straight line of an injection molded part, the DTW distances between the first and second regular sequences of the straight line and the central sequence are calculated respectively, and are recorded as the first and second DTW distances respectively; the mean of the color change characteristics of all pixel points on the straight line is calculated; the absolute value of the difference between the color change characteristics of each pixel point on the straight line and the mean is calculated; the sum of the absolute values ​​of the differences of all pixel points on the straight line is calculated; the product of the first DTW distance, the second DTW distance and the sum is calculated; and the calculation result of an exponential function with a natural constant as the base and the opposite number of the product as the exponent is used as the regularity of the gas mark of the abnormally extended straight line of the injection molded part.

[0024] Preferably, the overall quality of the injection molded parts is obtained according to the regularity of the gas marks of all abnormally extended straight lines of the injection molded parts, specifically:

[0025] When the regularity of the air mark of the abnormally extended straight line of the injection molded part is less than a preset regularity threshold, the abnormally extended straight line of the injection molded part is an air mark defect of the injection molded part; otherwise, the abnormally extended straight line of the injection molded part is the normal structure of the injection molded part; if the number of the air mark defects is greater than the preset number threshold, the overall quality of the injection molded part is poor; otherwise, the overall quality of the injection molded part is good.

[0026] The beneficial effects of the present invention are:

[0027] The present invention calculates the corresponding color change characteristics according to the color value change characteristics at different pixel positions in the HSV image of the injection molded part surface, and at the same time obtains different super-pixel segmentation image blocks and the shape center of the super-pixel segmentation image blocks according to the color change characteristics, which is more accurate than the super-pixel segmentation image blocks obtained only by the grayscale value of the pixel points; the local uniform feature coefficient is calculated by using the relative position between different pixel points and the shape center pixel point in different super-pixel segmentation image blocks to calculate and characterize the abnormal gas mark defect characteristics in the injection molded parts; further, the present invention obtains the abnormal image blocks of the injection molded parts according to the local uniform feature coefficients of different super-pixel segmentation image blocks, thereby narrowing the search range of abnormal gas marks; constructs the Hessian matrix of the abnormal image blocks of the injection molded parts to obtain the main extension direction of the gas mark defects; and constructs the gas mark regularity by the change of the color value characteristics of the pixel points under the main extension direction to clearly distinguish the abnormal gas mark characteristics on the surface of the thermoplastic elastomer injection molded parts, effectively avoiding the disadvantage of poor abnormal gas mark recognition accuracy caused by the inability of the traditional LSD algorithm to judge the extension direction of the abnormal gas mark, thereby improving the efficiency and accuracy of abnormal gas mark detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0029] Figure 1 A schematic flow chart of a method for identifying anomalies in injection molding of a thermoplastic elastomer provided by an embodiment of the present invention;

[0030] Figure 2 Schematic diagram of the steps of the method for identifying anomalies in injection molding of thermoplastic elastomers. DETAILED DESCRIPTION

[0031] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0032] See also Figure 1 , which shows a flow chart of a method for identifying anomalies in injection molding of a thermoplastic elastomer provided by an embodiment of the present invention, the method comprising the following steps:

[0033] Step S001, collecting an RGB image of the surface of a thermoplastic elastomer injection-molded part, and obtaining an HSV image of the surface of the injection-molded part.

[0034] It should be noted that, in order to facilitate the subsequent detection of abnormal gas streak defects on the surface of thermoplastic elastomer injection molded parts, the surface image of the injection molded parts is obtained by a CCD camera. Since the image obtained by the CCD camera has three different channels of RGB, in order to minimize or even eliminate the influence of environmental noise interference on the image quality of the injection molded part surface during the shooting and acquisition process, the three different channels of RGB on the surface of the injection molded part are first used as the input of the Gaussian filtering algorithm, and the image processed by the Gaussian filtering algorithm is converted into the HSV color space, and the converted image is recorded as the HSV image of the surface of the injection molded part, wherein the Gaussian filtering algorithm is a well-known technology, and the specific process is not repeated.

[0035] Step S002, obtaining the color change characteristics of each pixel point according to the color change in the pixel point neighborhood in the HSV image of the injection molded part surface; performing superpixel segmentation according to the color change characteristics, constructing a local uniform feature coefficient, and screening abnormal image blocks of the injection molded part.

[0036] It should be noted that due to the residual gas in the injection mold or the gas generated by the decomposition of the plastic melt, abnormal gas mark defects appear on the surface of the thermoplastic elastomer injection molded parts. At this time, the color of the abnormal gas mark defect position on the surface of the injection molded part will be different from the normal position on the surface of the injection molded part.

[0037] Normally, abnormal gas mark defects exist in local areas on the surface of injection molded parts. Therefore, in order to calculate the color change characteristics at the gas mark defect location on the surface of injection molded parts, the color change characteristics of each pixel in the HSV image of the injection molded part surface are constructed. As an example, the color change characteristics may include: constructing a neighborhood window area of ​​size W×W with each different pixel in the HSV image of the injection molded part surface as the center, assuming that W takes an empirical value of 3. In specific applications, the implementer can set it according to the specific situation. The expression for calculating the color change characteristics based on the color change of the pixel points in the neighborhood is:

[0038]

[0039] In the above formula, Cds i represents the color change characteristics at pixel i in the HSV image of the injection molded part surface, σ i and μ i They represent the variance and mean of the H channel value in the neighborhood window area centered on pixel i in the HSV image of the injection molded part surface, MaxH i and MinH i They respectively represent the maximum and minimum values ​​of the H channel in the neighborhood window area centered on pixel point i in the HSV image of the injection molded part surface.

[0040] The color change characteristics at different pixel positions in the HSV image of the injection molded part surface are calculated using the above formula.

[0041] For the neighborhood window area centered on pixel i in the HSV image of the injection molded part surface, the value of the H channel represents the color state at different pixel positions in the HSV image of the injection molded part surface. Therefore, in the neighborhood window area, if the range of the H channels of all pixels at different positions is larger, the coefficient of variation will also be larger. At this time, the calculated color change feature value will also be larger, indicating that the difference in the H channel values ​​of all different pixels in the neighborhood window area is also relatively large, and the possibility of abnormal gas mark defects occurring in the neighborhood window area of ​​the corresponding pixel is also relatively large.

[0042] Furthermore, in order to determine the area where the abnormal gas mark defect is located, superpixel segmentation is performed based on the color change characteristics of each pixel point, and the local uniform feature coefficient of each superpixel segmented image block is constructed according to the position and color change characteristics of each pixel point in each superpixel segmented image block. Based on the local uniform feature coefficient, the abnormal image block of the injection molded part in the superpixel segmented image block is obtained to narrow the area where the abnormal gas mark defect may exist. As an example, the acquisition of the abnormal image block of the injection molded part may include:

[0043] First, the color change feature at each pixel position is arranged according to the position of each pixel point on the surface of the injection molded part to obtain an image as the color change feature image of the HSV image of the injection molded part surface. The color change feature image is used as the input of the superpixel segmentation algorithm to obtain N different superpixel segmentation image blocks, wherein the superpixel segmentation algorithm is a well-known technology and the specific process is not repeated here.

[0044] Then, it should be noted that the number of pixels of different superpixel blocks obtained by the superpixel segmentation algorithm may be different. For the kth superpixel segmentation image block, the kth superpixel segmentation image block is divided into four different areas with equal numbers of pixels using horizontal and vertical straight lines, and the intersection of the horizontal and vertical straight lines is recorded as the shape center of the kth superpixel segmentation image block, where the horizontal and vertical straight lines are recorded as the segmentation lines of the kth superpixel segmentation image block; the expression for calculating the local uniform feature coefficient based on the color change characteristics of each pixel point in the superpixel segmentation image block and the distance to the shape center is:

[0045]

[0046] In the above formula, Stb k represents the local uniform feature coefficient of the kth superpixel image segmentation block, num k represents the total number of pixels in the kth superpixel segmentation image block, represents the color change feature of pixel i in the kth superpixel segmentation image block, Gaus(*) represents the Gaussian distribution function, It represents the Euclidean distance between pixel i and the center of the shape in the kth superpixel segmented image block.

[0047] The local uniform feature coefficients of different super-pixel image segments are calculated using the above formula.

[0048] In the different super-pixel segmentation image blocks obtained after division, if there is no abnormal gas mark defect on the surface of the injection molded part, the color change feature values ​​at different pixel positions in the super-pixel segmentation image block should not be much different. On the contrary, if a more obvious abnormal gas mark defect appears in the super-pixel segmentation image block obtained after division, the color change feature value difference of the pixel points of the entire super-pixel segmentation image block is relatively obvious. Therefore, for different super-pixel segmentation image blocks, the Gaussian distribution function value of the distance between different pixel points and the center pixel point of the body is taken as the weight to calculate the local uniform feature coefficient of different super-pixel segmentation image blocks.

[0049] Finally, the difference between the maximum and minimum values ​​in the color change characteristics of all super-pixel segmented image blocks is used as the judgment threshold of the abnormal gas pattern image block of the injection molded part. It should be noted that the implementer can also set the judgment threshold in other ways, such as calculating the threshold in the local uniform feature coefficient by the Otsu threshold algorithm, and using the threshold as the judgment threshold; calculating the difference between the local uniform feature coefficient of each super-pixel segmented image block and the color change feature of the center pixel of the shape, and recording it as the first difference. If the first difference is greater than or equal to the judgment threshold, it means that the color change feature of the center pixel of the shape of the super-pixel segmented image block is different from the numerical uniform change characteristics of the color change feature calculated at other pixel positions in the super-pixel segmented image block, and the current super-pixel segmented image block is likely to have abnormal gas pattern defects. Therefore, all different super-pixel segmented image blocks whose first difference is greater than or equal to the judgment threshold of the abnormal gas pattern image block of the injection molded part are obtained, and recorded as the abnormal image blocks of the injection molded part. It is assumed that there are Nd abnormal image blocks of the injection molded part.

[0050] Step S003, obtaining the main extension direction of the gas mark defect according to the Hessian matrix of the abnormal image block of the injection molded part, screening the abnormal extension straight line of the injection molded part in the abnormal image block of the injection molded part, and obtaining the gas mark regularity of the abnormal extension straight line of the injection molded part according to the changes of the pixel points on the straight line and the surrounding pixel points.

[0051] It should be noted that the abnormal gas mark defects on the surface of injection molded parts are usually caused by residual gas in the injection mold or gas generated by the decomposition of the plastic melt. The abnormal gas mark defects on the surface of injection molded parts have strong random and irregular change characteristics. This irregularly changing abnormal gas mark may overlap with the original texture on the surface of the injection molded parts. Therefore, it is necessary to combine Nd different abnormal image blocks of injection molded parts to further calculate the texture on the surface of the injection molded parts.

[0052] For the kth abnormal image block of injection molded parts, suppose it has num k pixel points, and these pixel points are arranged in order from top to bottom and from left to right; each pixel point can calculate the corresponding color change feature value, and the color change features of all pixel points are arranged in sequence according to the arrangement order of each pixel point to form a matrix as the injection molded parts abnormal change matrix. In order to facilitate subsequent further calculations, the injection molded parts abnormal change matrix is ​​expanded into a square matrix of size Z, where The function of rounding up is represented, and the mean value of the color change characteristic values ​​of all pixels in the abnormal image block of the injection molded part is used for expansion, so that the abnormal change matrix of the injection molded part is complete. The abnormal change matrix of the injection molded part is used as the input of the Hessian matrix calculation to obtain the Hessian matrix of the abnormal change matrix of the injection molded part, which is recorded as the Hessian matrix of the abnormal change of the injection molded part; the Hessian matrix of the abnormal change of the injection molded part is used as the input of the eigenvalue decomposition algorithm to obtain the eigenvalues ​​of the Hessian matrix of the abnormal change of the injection molded part, wherein the construction process of the Hessian matrix and the specific calculation process of the eigenvalue decomposition algorithm are well-known technologies and will not be repeated here.

[0053] Among the multiple different eigenvalues ​​obtained after the Hessian matrix of abnormal changes in injection molded parts is subjected to the eigenvalue decomposition algorithm, the eigenvector corresponding to the maximum eigenvalue reflects the direction in which the color change characteristic value of the pixel point in the abnormal image block of the injection molded part changes greatly, while the eigenvector corresponding to the minimum eigenvalue reflects the direction in which the color change characteristic value of the pixel point in the abnormal image block of the injection molded part changes less. When abnormal gas pattern defects appear on the surface of the injection molded part, the difference in the color change characteristic value on both sides of the abnormal gas pattern will be relatively obvious, while in the extension direction of the abnormal gas pattern, the difference in the color change characteristic value of the abnormal gas pattern at different pixel positions is relatively small. Therefore, the direction of the eigenvector corresponding to the minimum eigenvalue of the Hessian matrix of the abnormal change of the injection molded part of the kth abnormal image block of the injection molded part is taken as the main extension direction of the gas pattern defect of the kth abnormal image block of the injection molded part.

[0054] The main extension direction of the gas mark defect in each abnormal image block of the injection molded part is obtained by acquiring the main extension direction of the gas mark defect in the kth abnormal image block of the injection molded part.

[0055] It should be noted that the most obvious distinguishing feature between abnormal air mark defects and other inherent lines on the surface of thermoplastic elastomer injection molded bodies is the irregular change characteristics of the abnormal air mark defects. Therefore, the change characteristics of the lines in the abnormal image blocks of different injection molded parts are analyzed and calculated based on the main extension direction of the abnormal air mark defects of the injection molded parts.

[0056] The k-th abnormal image block of the injection molded part is used as the input of the LSD line detection algorithm, all different straight lines in the k-th abnormal image block of the injection molded part are obtained, and all different straight lines are screened using their corresponding main extension directions, and all straight lines in the k-th abnormal image block of the injection molded part with the same main extension direction are obtained, and are recorded as abnormal extension straight lines of the injection molded part. Among them, the LSD line detection algorithm is a well-known technology, and the specific process is not repeated here.

[0057] The abnormal extension straight line of the injection molded part in each abnormal image block of the injection molded part is obtained by acquiring the abnormal extension straight line of the injection molded part in the kth abnormal image block of the injection molded part.

[0058] For each abnormal extension line of each injection molded part, the color change characteristics of all pixels on the abnormal extension line of the injection molded part are arranged in the order of the pixel coordinate size in the main extension direction to obtain the center sequence of the line, and the color change characteristics of the pixels on both sides of the line are arranged in the order of the pixel coordinate size in the main extension direction to obtain the first regular sequence and the second regular sequence of the line. The expression for the regularity of the gas pattern of each abnormal extension line of each injection molded part is constructed based on the center sequence and the first and second regular sequences of each abnormal extension line of each injection molded part:

[0059]

[0060] In the formula, SLm t represents the regularity of the gas pattern of the tth abnormal extension line of the injection molded part, exp[*] represents the exponential function with the natural constant as the base, Nt represents the total number of pixels in the tth abnormal extension line of the injection molded part, Cds j It represents the color change feature at the jth pixel position in the abnormal extension line of the tth injection molded part, M t represents the mean value of the color change characteristics of all different pixels in the abnormal extension line of the t-th injection molded part. DTW(*) represents the DTW distance between two different sequences. Lg t,1 , Lg t.c and Lg t,2 They represent the first regular sequence, central sequence and second regular sequence of the abnormal extension line of the tth injection molded part. Among them, DTW(Lg t,1 ,Lg t.c ) is the first DTW distance, DTW(Lg t,2 ,Lg t,c ) is the second DTW distance.

[0061] The above formula can be used to calculate the regularity of the gas pattern of the abnormal extension line of the tth injection molded part. Assuming that the difference between the color change characteristics and the mean at different pixel positions on the abnormal extension line of the tth injection molded part is smaller, and the DTW distance between the first regular sequence and the second regular sequence on both sides of the abnormal extension line of the tth injection molded part is smaller, it means that the difference in the color change characteristics on the abnormal extension line of the tth injection molded part and on both sides is relatively small. At this time, the gas pattern regularity value calculated at the corresponding straight line is relatively large, indicating that the overall change of the abnormal extension line of the tth injection molded part is relatively regular. On the contrary, if the calculated gas pattern regularity value is relatively small, it means that the overall change of the straight line at the current position is irregular, and it is likely to represent an abnormal gas pattern defect on the surface of the injection molded part.

[0062] Step S004, judging the overall quality of the injection molded part according to the regularity of the air marks.

[0063] A regularity threshold is set. It should be noted that the regularity threshold can be set by the implementer. In this embodiment, the regularity threshold is set to 0.5. If the regularity of the air marks of each abnormally extended straight line of the injection molded part is less than the regularity threshold, the abnormally extended straight line of the injection molded part is considered to be an air mark defect on the injection molded part; if the regularity of the air marks of each abnormally extended straight line of the injection molded part is greater than or equal to the regularity threshold, the abnormally extended straight line of the injection molded part is a normal structure of the injection molded part. The schematic diagram of the steps of the above method is shown in FIG. Figure 2 As shown. It should be noted that the quantity threshold is set by the implementer. In this embodiment, the quantity threshold is set to 3. If the number of gas mark defects obtained by the above regularity threshold detection is greater than the quantity threshold, it is considered that the overall quality of the injection molded part is poor, otherwise it is considered that the overall quality of the injection molded part is good.

[0064] The embodiments described above are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, a person skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the scope of the technical solutions of the embodiments of the present application, and should all be included in the protection scope of the present application.

Claims

1. A method for identifying anomalies in injection molding of a thermoplastic elastomer, characterized in that: The method comprises the following steps: Collect RGB images of the surface of the injection molded parts and obtain HSV images of the RGB images; In the HSV image of the surface of the injection molded part, the color change characteristics of each pixel are obtained according to the change of the H channel value in the neighborhood of each pixel; each superpixel segmentation image block is obtained according to the color change characteristics of each pixel; the local uniform characteristic coefficient of each superpixel segmentation image block is obtained according to the difference in the color change characteristics of the pixel; the abnormal image block of the injection molded part in the superpixel segmentation image block is obtained by screening according to the local uniform characteristic coefficient; The main extension direction of the gas mark defect in the abnormal image block of the injection molded part is obtained according to the color change characteristics of the pixel points in the abnormal image block of the injection molded part; the abnormal extension straight lines of each injection molded part in the abnormal image block of the injection molded part are obtained according to the main extension direction of the gas mark defect and the LSD straight line detection algorithm; the gas mark regularity of each abnormal extension straight line of the injection molded part is obtained according to the difference in color change characteristics between the pixel points on each abnormal extension straight line of the injection molded part and the surrounding pixel points; The overall quality of the injection molded parts can be obtained based on the regularity of the air marks on the abnormally extended straight lines of all injection molded parts; The main extension direction of the gas mark defect in the abnormal image block of the injection molded part is obtained according to the color change characteristics of the pixel points in the abnormal image block of the injection molded part, specifically: A matrix composed of color change features of all pixels in the abnormal image block of each injection molded part is used as the abnormal change matrix of each injection molded part, wherein the empty values ​​in the abnormal change matrix of the injection molded parts are filled with the mean value of the color change feature values ​​of all pixels in the abnormal image block of the injection molded parts; the Hessian matrix of the abnormal change matrix of each injection molded part is obtained; the different eigenvalues ​​of each Hessian matrix are obtained by the eigenvalue decomposition algorithm; the direction of the eigenvector corresponding to the minimum eigenvalue of each Hessian matrix is ​​used as the main extension direction of the air mark defect of the abnormal image block of the injection molded part.

2. A method for identifying anomalies in injection molding of a thermoplastic elastomer according to claim 1, characterized in that: The color change feature of each pixel point is obtained according to the change of the H channel value in the neighborhood of each pixel point, specifically including: For the neighborhood of pixel point i, calculate the mean and variance of all H channel values ​​in the neighborhood; calculate the absolute value of the difference between the maximum and minimum values ​​of the H channel values ​​in the neighborhood; calculate the ratio of the variance to the mean; and use the product of the ratio and the absolute value of the difference as the color change feature of pixel point i.

3. A method for identifying anomalies in injection molding of a thermoplastic elastomer according to claim 1, characterized in that: The method of obtaining each superpixel segmented image block according to the color change characteristics of each pixel point is specifically as follows: an image composed of the color change characteristics of all pixels is used as a color change characteristic image; and the color change characteristic image is subjected to superpixel segmentation by a superpixel segmentation algorithm to obtain each superpixel segmented image block.

4. A method for identifying anomalies in injection molding of a thermoplastic elastomer according to claim 1, characterized in that: The method of obtaining the local uniform characteristic coefficient of each super-pixel segmented image block according to the color change characteristic difference of the pixel points specifically includes: Obtain the intersection of the segmentation lines that divide the area of ​​the superpixel segmented image block into equal parts; in each superpixel segmented image block, calculate the Euclidean distance between each pixel point and the intersection point; calculate the product of the value of the Gaussian distribution function of the Euclidean distance and the color change feature of each pixel point; and use the mean value of the product of all pixels as the local uniform feature coefficient of the superpixel segmented image block.

5. A method for identifying anomalies in injection molding of a thermoplastic elastomer according to claim 1, characterized in that: The abnormal image blocks of the injection molded parts in the superpixel segmented image blocks are obtained by screening according to the local uniform feature coefficient, specifically: The difference between the maximum and minimum values ​​in the color change characteristics of all super-pixel segmented image blocks is used as the judgment threshold; the difference between the local uniform feature coefficient of each super-pixel segmented image block and the color change characteristics of the central pixel point of the shape is calculated; the super-pixel segmented image block with the difference greater than or equal to the judgment threshold is used as the abnormal image block of the injection molded part.

6. A method for identifying anomalies in injection molding of a thermoplastic elastomer according to claim 1, characterized in that: The method of obtaining abnormal extension straight lines of each injection molded part in the abnormal image block of the injection molded part according to the main extension direction of the air mark defect and the LSD straight line detection algorithm is specifically as follows: each straight line in each abnormal image block of the injection molded part is obtained by the LSD straight line detection algorithm; and the straight line in each abnormal image block of the injection molded part having the same direction as the main extension direction is used as the abnormal extension straight line of each injection molded part in the abnormal image block of each injection molded part.

7. A method for identifying anomalies in injection molding of a thermoplastic elastomer according to claim 1, characterized in that: The gas pattern regularity of the abnormal extension line of each injection molded part is obtained according to the color change characteristic difference between the pixel point on the abnormal extension line of each injection molded part and the surrounding pixel points, specifically: The sequence composed of the color change characteristics of all pixel points on the abnormal extension straight line of each injection molded part is taken as the central sequence, and the sequence composed of the color change characteristics of the pixel points on both sides of the straight line is recorded as the first and second regular sequences respectively; the regularity of the air marks on the abnormal extension straight line of each injection molded part is obtained according to the difference between the first and second regular sequences and the central sequence.

8. A method for identifying anomalies in injection molding of a thermoplastic elastomer according to claim 7, characterized in that: The method of obtaining the regularity of the air marks of the abnormally extended straight lines of each injection molded part according to the difference between the first and second regular sequences and the central sequence specifically includes: For each abnormally extended straight line of an injection molded part, the DTW distances between the first and second regular sequences of the straight line and the central sequence are calculated respectively, and are recorded as the first and second DTW distances respectively; the mean of the color change characteristics of all pixel points on the straight line is calculated; the absolute value of the difference between the color change characteristics of each pixel point on the straight line and the mean is calculated; the sum of the absolute values ​​of the differences of all pixel points on the straight line is calculated; the product of the first DTW distance, the second DTW distance and the sum is calculated; and the calculation result of an exponential function with a natural constant as the base and the opposite number of the product as the exponent is used as the regularity of the gas mark of the abnormally extended straight line of the injection molded part.

9. A method for identifying anomalies in injection molding of a thermoplastic elastomer according to claim 1, characterized in that: The overall quality of the injection molded parts is obtained according to the regularity of the air marks of the abnormally extended straight lines of all the injection molded parts, specifically: When the regularity of the air mark of the abnormally extended straight line of the injection molded part is less than a preset regularity threshold, the abnormally extended straight line of the injection molded part is an air mark defect of the injection molded part; otherwise, the abnormally extended straight line of the injection molded part is the normal structure of the injection molded part; if the number of the air mark defects is greater than the preset number threshold, the overall quality of the injection molded part is poor; otherwise, the overall quality of the injection molded part is good.

Images