Semiconductor ceramic composite material surface defect detection method based on image processing

Abstract

The invention relates to the technical field of defect detection, in particular to a semiconductor ceramic composite material surface defect detection method based on image processing. Obtaining a local gradient size difference value and a local gradient direction difference value according to gradient difference features of adjacent edge pixel points in the target edge; obtaining an edge smoothing characteristic value according to the distribution characteristics of the local gradient size difference value and the local gradient direction difference value; obtaining a smoothness difference degree according to the edge smoothness feature values of any two target edges; obtaining a defect probability value according to the smooth difference degree and judging whether a defect exists or not; and clustering the target edge according to the smooth difference degree, and selecting a defect edge cluster according to the edge smooth feature value. The method comprises the following steps: obtaining a seed point and a defect region of a watershed segmentation algorithm according to gray difference characteristics of regions on two sides of a defect edge in a defect edge cluster; the defect degree is obtained according to the area characteristics of the defect region, and the accuracy of seed point selection and defect detection is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of defect detection, and specifically relates to a method for detecting surface defects of semiconductor ceramic composites based on image processing. Background Art

[0002] Semiconductor ceramic composites are composite materials composed of ceramic matrices and semiconductor materials, and are widely used in fields such as electronic packaging, sensors, and power devices due to their excellent properties. Due to the brittle characteristics of this material, during the preparation process of this material, defects such as peeling and cracks are likely to appear on the surface, affecting product quality; therefore, surface defect detection is required. Through existing machine vision technology, automated detection can be carried out. The existing watershed segmentation algorithm can segment and extract the defect intervals on the surface of the material. The accuracy of the defect area segmentation selected by this algorithm depends on the selection of seed points, and the selection of seed points depends on the gray distribution characteristics of the image; however, due to the three-dimensional structure of the ceramic composite material, the colors at different positions on the surface are different in brightness due to the influence of the shooting light, making the colors of the defect surface and some normal areas close, thus affecting the selection of seed points, and ultimately resulting in a low segmentation accuracy of the watershed segmentation algorithm for the surface defect area, and ultimately affecting the accuracy of defect detection. Summary of the Invention

[0003] In order to solve the above technical problem that the colors of the defect surface and some normal areas are close due to the shooting light, resulting in a low segmentation accuracy of the watershed segmentation algorithm for the surface defect area and affecting the accuracy of defect detection, the purpose of the present invention is to provide a method for detecting surface defects of semiconductor ceramic composites based on image processing, and the specific technical solution adopted is as follows: Obtain the appearance image of the semiconductor ceramic composite material; Obtain edge intersection points according to the quantity characteristics of other edge pixels within the neighborhood range of the edge pixels on the upper edge of the edge line in the appearance image; obtain different target edges in the edge line according to the edge intersection points; obtain the local gradient magnitude difference value and the local gradient direction difference value according to the gradient difference characteristics of adjacent edge pixels in the target edges; Obtain the edge smoothness eigenvalue according to the distribution characteristics of the local gradient magnitude difference value and the local gradient direction difference value of the edge pixels in the target edges; obtain the smoothness difference degree according to the difference characteristics of the edge smoothness eigenvalues of any two target edges; obtain the defect probability value according to the discrete characteristics of the smoothness difference degree and determine whether there are defects; if there are defects, cluster the target edges according to the smoothness difference degree, and select the defect edge cluster in the clustering result according to the edge smoothness eigenvalue; Obtain the seed points of the watershed segmentation algorithm according to the gray-scale difference characteristics of the two-side regions of any defect edge in the defect edge cluster; obtain the defect region of the composite material appearance through the watershed segmentation algorithm based on the seed points; obtain the defect degree according to the area characteristics of the defect region.

[0004] Further, the step of obtaining the edge intersection points according to the quantity characteristics of other edge pixel points within the neighborhood range of the edge pixel points on the edge line in the appearance image includes: When there are no less than 3 other edge pixel points in the eight-neighborhood range of any edge pixel point, the any edge pixel point is an edge intersection point.

[0005] Further, the step of obtaining different target edges in the edge line according to the edge intersection points includes: Taking any edge intersection point as the starting point, use the local edge line between the any edge intersection point and the other directly-connected edge intersection points as the target edge.

[0006] Further, the step of obtaining the local gradient magnitude difference value and the local gradient direction difference value according to the gradient difference characteristics of adjacent edge pixel points in the target edge includes: , take the gradient magnitude and gradient direction of the edge pixel points in the target edge as the gradient feature values respectively; in the formula, represents the gradient feature value of the i-th edge pixel point in the target edge, represents the -th edge pixel point's gradient feature value, represents the -th edge pixel point's gradient feature value; represents the difference between the gradient feature values of the -th and the -th edge pixel points, represents the difference between the gradient feature values of the i-th and the -th edge pixel points, represents the difference between the gradient feature values of the i-th and the -th edge pixel points; when the gradient feature value represents the gradient magnitude, represents the local gradient magnitude difference value of the i-th edge pixel point, when the gradient feature value represents the gradient direction, represents the local gradient direction difference value of the i-th edge pixel point.

[0007] Further, the step of obtaining the edge smoothness feature value according to the distribution characteristics of the local gradient magnitude difference value and the local gradient direction difference value of the edge pixel points in the target edge includes: Calculate the standard deviation of the difference values of the local gradient magnitudes of the edge pixel points in the target edge to obtain the first distribution disorder degree; calculate the standard deviation of the difference values of the local gradient directions of the edge pixel points in the target edge to obtain the second distribution disorder degree; calculate the sum value of the first distribution disorder degree and the second distribution disorder degree and perform a negative correlation mapping to obtain the edge smoothness feature value of the target edge.

[0008] Further, the step of obtaining the smoothness difference degree according to the difference features of the edge smoothness feature values of any two target edges includes: Calculate the absolute value of the difference between the edge smoothness feature values of any two target edges to obtain the smoothness difference degree between the any two target edges.

[0009] Further, the step of obtaining the defect probability value according to the discrete feature of the smoothness difference degree and determining whether there is a defect includes: Calculate the difference between the maximum value and the minimum value of the smoothness difference degree and normalize it to obtain the defect probability value of the appearance of the composite material; when the defect probability value does not exceed the preset normal threshold, there is a defect in the appearance of the composite material.

[0010] Further, the step of clustering the target edges according to the smoothness difference degree and selecting the defective edge cluster from the clustering result according to the edge smoothness feature value includes: Cluster according to the smoothness difference degree between the target edges by the K-means algorithm to obtain two target edge clusters; take the target edge cluster containing the minimum value of the edge smoothness feature value as the defective edge cluster.

[0011] Further, the step of obtaining the seed points of the watershed segmentation algorithm according to the gray level difference feature of the two side regions of any defective edge in the defective edge cluster includes: Taking the middle edge pixel point of the any defective edge as the starting point, taking a preset number of connected pixel points along the gradient direction of the middle edge pixel point to construct a pixel point sequence; taking a preset number of connected pixel points along the opposite direction of the gradient direction of the middle edge pixel point to construct another pixel point sequence; performing a difference on the gray level values of the pixel point sequences and calculating the standard deviation to obtain the gray level disorder degree; taking any pixel point in the pixel point sequence corresponding to the maximum value of the gray level disorder degree as the seed point.

[0012] Further, the step of obtaining the defect degree according to the area feature of the defective region includes: Calculate the ratio of the number of pixel points of the defective region to the number of pixel points of the appearance image to obtain the defect degree.

[0013] The present invention has the following beneficial effects: In the embodiments of the present invention, obtaining edge intersection points can split the initially obtained edge lines, so that each target edge only represents a separate feature on the surface of the composite material, improving the accuracy of subsequent defect segmentation. Obtaining the local gradient magnitude difference value and local gradient direction difference value of edge pixel points can characterize the gradient difference features at the edge pixel points. Since there are differences in the smooth regularity of the gradient features of the edge lines in the defect area and the normal area, obtaining the edge smoothness feature value can characterize the possibility that the target edge is a defect area. Obtaining the smoothness difference can characterize the difference in the edge smoothness degree between different target edges. Furthermore, obtaining the defect probability value based on the smoothness difference can determine the probability of the existence of a defect area on the surface of the composite material. Obtaining the defect edge cluster can determine the target edge position of the defect area. Obtaining the seed points of the watershed segmentation algorithm based on the gray-scale difference features of the areas on both sides of any defect edge in the defect edge cluster can accurately segment the defect area, avoiding errors in the defect area segmentation caused by determining the seed points based on the gray-scale values, thereby improving the accuracy of the surface defect detection of the composite material. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the following-described drawings are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0015] Figure 1 Flowchart of a method for detecting surface defects of a semiconductor ceramic composite material based on image processing provided by an embodiment of the present invention; Figure 2 Appearance image of a semiconductor ceramic composite material provided by an embodiment of the present invention; Figure 3 Edge line image of a semiconductor ceramic composite material provided by an embodiment of the present invention; Figure 4 Defect area image of a semiconductor ceramic composite material provided by an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0016] To further elaborate on the technical means and effects adopted by the present invention to achieve the intended invention purpose, the following specifically describes, in conjunction with the accompanying drawings and preferred embodiments, a method for detecting surface defects of a semiconductor ceramic composite material based on image processing, including its specific implementation manner, structure, features, and effects. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. In addition, the specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0017] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the technical field to which the present invention belongs.

[0018] The following specifically describes, in conjunction with the accompanying drawings, the specific solution of a method for detecting surface defects of a semiconductor ceramic composite material based on image processing provided by the present invention.

[0019] Please refer to Figure 1 , which shows a flowchart of a method for detecting surface defects of a semiconductor ceramic composite material based on image processing provided by an embodiment of the present invention. The method includes the following steps: Step S1, obtaining an appearance image of the semiconductor ceramic composite material.

[0020] In an embodiment of the present invention, the implementation scenario is to detect surface defects of a semiconductor ceramic composite material through image processing technology to improve the accuracy of defect detection. First, an appearance image of the semiconductor ceramic composite material is obtained. The semiconductor ceramic composite material in the embodiment of the present invention is a silicon nitride ceramic composite material. Figure 2 shows an appearance image of a semiconductor ceramic composite material.

[0021] Step S2, obtaining edge intersection points according to the quantity characteristics of other edge pixels within the neighborhood range of the upper-edge pixels of the edge line in the appearance image; obtaining different target edges in the edge line according to the edge intersection points; obtaining the local gradient magnitude difference value and the local gradient direction difference value according to the gradient difference characteristics of adjacent edge pixels in the target edges.

[0022] Under normal circumstances, the obtained appearance image can be subjected to defect segmentation through existing watershed segmentation algorithms. This algorithm selects appropriate seed point positions based on the gray-scale features in the image, and then uses the principle of rising or falling water to segment and extract the defect areas at the positions of the seed points. For example, if the color of the defect area is darker and the gray-scale value is lower, the pixel point with the smallest gray-scale value can be selected as the seed point for segmenting and extracting the defect area. However, in the current scenario, the ceramic composite material has a three-dimensional structure and a relatively complex surface structure. During the shooting process, due to the influence of light, the color features of the defect area and some normal areas are likely to be relatively close, thus affecting the accuracy of seed point selection and defect area segmentation. Therefore, in order to improve the accuracy of defect detection, it is necessary to improve the method of selecting seed points.

[0023] First, edge intersection points are obtained according to the quantity characteristics of other edge pixels within the neighborhood range of the edge pixels on the edge line in the appearance image; the edge line of the appearance image is obtained through existing edge detection algorithms, and the specific steps will not be elaborated here. Figure 3 Fig. shows an edge line image of a semiconductor ceramic composite material; different surface areas can be segmented through the edge line. According to Figure 3 It can be seen that a relatively long edge line may represent the edges of multiple planes in the composite material. Therefore, it is necessary to split the edge line to improve the accuracy of subsequent defect area recognition. Furthermore, edge intersection points are obtained according to the quantity characteristics of other edge pixels within the neighborhood range of the edge pixels on the edge line in the appearance image; preferably, in the embodiment of the present invention, the step of obtaining edge intersection points includes: when no less than 3 other edge pixels appear within the eight-neighborhood range of any edge pixel, it means that at least two edge lines intersect at the position of the any edge pixel, and the any edge pixel is an edge intersection point.

[0024] Further, after obtaining the intersection points of the edge lines, different target edges in the edge lines can be obtained based on the edge intersection points; preferably, in the embodiments of the present invention, the steps of obtaining the target edges include: taking any edge intersection point as a starting point, and using the local edge line between the any edge intersection point and other directly connected edge intersection points as the target edge. Each target edge can represent a separate feature segment on the surface of the composite material, making the determination of the defect range more accurate; it should be noted that if an edge line forms a closed connected domain and no edge intersection points appear, then this edge line is used as a target edge. When there are defects on the surface of the composite material, the shape of the defect edges is relatively random and uneven, while the edge lines of the normal area have strong smoothness and regularity; furthermore, reflected in the gradient characteristics of the edge pixel points, the regularity of the gradient characteristics of the defect edge pixel points is poor, while the gradient characteristics of the normal area edge pixel points are more regular. Therefore, the local gradient magnitude difference value and the local gradient direction difference value can be obtained according to the gradient difference characteristics of adjacent edge pixel points in the target edge. Preferably, in the embodiments of the present invention, the steps and formulas for obtaining the local gradient magnitude difference value and the local gradient direction difference value include:

[0025] Taking the gradient amplitude and gradient direction of the edge pixel points in the target edge as gradient feature values respectively; where, represents the gradient feature value of the i-th edge pixel point in the target edge, represents the -th edge pixel point's gradient feature value, represents the -th edge pixel point's gradient feature value; represents the difference between the gradient feature values of the -th and the -th edge pixel points, represents the difference between the gradient feature values of the i-th and the -th edge pixel points, represents the difference between the gradient feature values of the i-th and the -th edge pixel points; when the gradient feature value represents the gradient amplitude, represents the local gradient magnitude difference value of the i-th edge pixel point, and when the gradient feature value represents the gradient direction, represents the local gradient direction difference value of the i-th edge pixel point. When the difference between the gradient feature values of adjacent edge pixel points in the target edge is larger, then and have larger values, has a larger value; conversely, when the difference between the gradient feature values is smaller, then has a smaller value.

[0026] Step S3: Obtain the edge smoothness feature value according to the distribution characteristics of the local gradient magnitude difference value and the local gradient direction difference value of the edge pixel points in the target edge; obtain the smoothness difference degree according to the difference characteristics of the edge smoothness feature values of any two target edges; obtain the defect probability value according to the discrete characteristics of the smoothness difference degree and determine whether there is a defect; if there is a defect, cluster the target edges according to the smoothness difference degree, and select the defective edge cluster from the clustering results according to the edge smoothness feature value.

[0027] The local gradient magnitude difference value and the local gradient direction difference value of the edge pixel points reflect the local gradient difference characteristics of the edge pixel points. If the target edge is the edge line of the normal region, the gradient characteristics of the edge pixel points are more regular, and the local magnitude difference value and the local gradient direction difference value of each edge pixel point in the target edge are more similar; if the target edge is the edge line of the defective region, the gradient characteristics of the edge pixel points are more random, and the distribution of the local magnitude difference value and the local gradient direction difference value of each edge pixel point in the target edge is more discrete. Therefore, the edge smoothness feature value can be obtained according to the distribution characteristics of the local gradient magnitude difference value and the local gradient direction difference value of the edge pixel points in the target edge.

[0028] Preferably, in the embodiment of the present invention, the step of obtaining the edge smoothness feature value includes: calculating the standard deviation of the local gradient magnitude difference value of the edge pixel points in the target edge to obtain the first distribution chaos degree; calculating the standard deviation of the local gradient direction difference value of the edge pixel points in the target edge to obtain the second distribution chaos degree; calculating the sum value of the first distribution chaos degree and the second distribution chaos degree and performing a negative correlation mapping to obtain the edge smoothness feature value of the target edge. When the first distribution chaos degree and the second distribution chaos degree are smaller, it means that the values of the local gradient magnitude difference value and the local gradient direction difference value of the edge pixel points in the target edge are closer, and the greater the possibility that the target edge is the edge line of the normal region, and the greater the value of the edge smoothness feature value. When the first distribution chaos degree and the second distribution chaos degree are larger, it means that the value distributions of the local gradient magnitude difference value and the local gradient direction difference value of the edge pixel points in the target edge are more discrete, and the greater the possibility that the target edge is the edge line of the defective region, and the smaller the value of the edge smoothness feature value.

[0029] Furthermore, if there are defect areas on the surface of the composite material, the difference in the edge smoothness eigenvalues of the edge lines between the normal area and the defect area is relatively large. Therefore, the smoothness difference degree can be obtained according to the difference characteristics of the edge smoothness eigenvalues of any two target edges; preferably, in the embodiment of the present invention, the steps of obtaining the smoothness difference degree include: calculating the absolute value of the difference between the edge smoothness eigenvalues of any two target edges to obtain the smoothness difference degree between the two target edges. The larger the smoothness difference degree, the more likely it means that there is a normal area edge line and a defect area edge line among the two target edges. After obtaining the smoothness difference degree, the defect probability value can be obtained according to the discrete characteristics of the smoothness difference degree and it can be determined whether there are defects; preferably, in the embodiment of the present invention, the steps of obtaining the defect probability value and determining whether there are defects include: calculating the difference between the maximum value and the minimum value of the smoothness difference degree and normalizing it to obtain the defect probability value of the appearance of the composite material; when the defect probability value does not exceed the preset normal threshold, there are defects in the appearance of the composite material. When there are no defect areas in the composite material, the smoothness difference degree is relatively small, and the defect probability value is relatively small; the larger the defect probability value, the greater the probability that there are defect areas in the appearance of the composite material. In the embodiment of the present invention, the preset normal threshold is the average value of the defect probability values of 10 identical normal composite materials, and the implementer can determine it according to the implementation scenario by himself / herself.

[0030] When there are defects, it is necessary to detect the defect areas and analyze the defect degree. The smoothness difference degree between the target edges in the defect area is relatively small, and the smoothness difference degree between the target edges of the defect area and the normal area is relatively large; therefore, the target edges can be clustered according to the smoothness difference degree, and the defect edge cluster can be selected from the clustering results according to the edge smoothness eigenvalues; preferably, in the embodiment of the present invention, clustering is performed by the K-means algorithm according to the smoothness difference degree between the target edges to obtain two target edge clusters; it should be noted that the K-means algorithm belongs to the existing clustering technology, and the specific steps will not be elaborated. The smoothness difference degree is used as the measurement distance between the target edges during the clustering process. There is a target edge of the normal area and a target edge of the defect area in the two target edge clusters. The edge smoothness eigenvalue of the target edge of the defect area is relatively small. Therefore, the target edge cluster containing the minimum value of the edge smoothness eigenvalue is used as the defect edge cluster.

[0031] Step S4, obtaining the seed points of the watershed segmentation algorithm according to the gray difference characteristics of the areas on both sides of any defect edge in the defect edge cluster; obtaining the defect area of the appearance of the composite material through the watershed segmentation algorithm according to the seed points; obtaining the defect degree according to the area characteristics of the defect area.

[0032] After obtaining the defective edge clusters, one side of the defective edge is the defective area. Due to the peeling of the coating in the defective area, the regularity of the gray-scale change is weak, while the gray-scale characteristics of the normal area are more regular. Furthermore, the seed points of the watershed segmentation algorithm can be obtained according to the gray-scale difference characteristics of the areas on both sides of any defective edge in the defective edge clusters; preferably, in the embodiments of the present invention, the steps of obtaining the seed points include: taking a preset number of connected pixel points along the gradient direction of the middle edge pixel point of any defective edge to construct a pixel point sequence; taking a preset number of connected pixel points along the opposite direction of the gradient direction of the middle edge pixel point to construct another pixel point sequence; in the embodiments of the present invention, the preset number is 5, and the implementer can determine it according to the implementation scenario. The gray-scale values of the pixel point sequence are differentiated and the standard deviation is calculated to obtain the gray-scale chaos degree; when the gray-scale values of the pixel point sequence are relatively similar or have regular changes, the standard deviation after gray-scale differentiation is small and the gray-scale chaos degree is small; when the gray-scale values of the pixel point sequence have changes and the changes are irregular, the standard deviation after gray-scale differentiation is large and the gray-scale chaos degree is large. Therefore, the side where the pixel point sequence with a larger gray-scale chaos degree is located is the defective area, and then any pixel point in the pixel point sequence corresponding to the maximum value of the gray-scale chaos degree is used as the seed point.

[0033] Further, after obtaining the seed points, the defective area of the composite material appearance can be obtained by the watershed segmentation algorithm according to the seed points; starting from the seed points until the closed edge line around the seed points is reached, a complete defective area is obtained. It should be noted that the watershed segmentation algorithm belongs to the prior art, and the specific steps will not be elaborated here; Figure 4 An image of the defective area of a semiconductor ceramic composite material is shown. By using this method to determine the seed points and obtain the defective area, it is possible to avoid the influence of the shooting light on the gray-scale values of the surface pixel points of the composite material, resulting in inaccurate selection of the seed points, thereby improving the accuracy of the segmentation and extraction of the defective area. After obtaining all the defective areas, the defective degree can be obtained according to the area characteristics of the defective areas; preferably, in the embodiments of the present invention, the steps of obtaining the defective degree include: calculating the ratio of the number of pixel points in the defective area to the number of pixel points in the appearance image to obtain the defective degree. When the area of the defective area is larger and the number of pixel points is more, the defective degree is larger. Subsequently, the composite material can be selected according to the defective degree.

[0034] In summary, the embodiment of the present invention provides a method for detecting surface defects of semiconductor ceramic composites based on image processing; obtaining local gradient magnitude difference values and local gradient direction difference values according to the gradient difference characteristics of adjacent edge pixel points in the target edge; obtaining edge smoothness eigenvalue according to the distribution characteristics of local gradient magnitude difference values and local gradient direction difference values; obtaining smoothness difference degree according to the edge smoothness eigenvalue of any two target edges; obtaining defect probability values according to the smoothness difference degree and determining whether there are defects; clustering the target edges according to the smoothness difference degree and selecting defect edge clusters according to the edge smoothness eigenvalue. The present invention obtains seed points and defect regions of the watershed segmentation algorithm according to the gray level difference characteristics of the two regions on both sides of the defect edge in the defect edge cluster; obtaining the defect degree according to the area characteristics of the defect region, which improves the accuracy of seed point selection and defect detection.

[0035] It should be noted that the above sequence of embodiments of the present invention is only for description and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the accompanying drawings do not necessarily require the particular order or sequential order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0036] Each embodiment in this specification is described in a progressive manner, and the same or similar parts among the embodiments can be referred to each other. Each embodiment focuses on the differences from other embodiments.

Claims

1. A method for detecting surface defects of semiconductor ceramic composite materials based on image processing, characterized in that: The method comprises the following steps: Obtain appearance images of semiconductor ceramic composite materials; Obtain edge intersection points according to the quantity characteristics of other edge pixels in the neighborhood of the edge pixel point on the edge line in the appearance image; obtain different target edges in the edge line according to the edge intersection points; obtain local gradient magnitude difference values ​​and local gradient direction difference values ​​according to the gradient difference characteristics of adjacent edge pixels in the target edge; Obtain edge smoothing feature values ​​according to the distribution characteristics of local gradient magnitude difference values ​​and local gradient direction difference values ​​of edge pixels in the target edge; obtain smoothing difference according to the difference characteristics of edge smoothing feature values ​​of any two target edges; obtain defect probability values ​​according to the discrete characteristics of the smoothing difference and judge whether there is a defect; if there is a defect, cluster the target edge according to the smoothing difference and select the defect edge cluster from the clustering result according to the edge smoothing feature value; The seed point of the watershed segmentation algorithm is obtained according to the grayscale difference characteristics of the areas on both sides of any defect edge in the defect edge cluster; the defect area of ​​the composite material appearance is obtained by the watershed segmentation algorithm according to the seed point; and the defect degree is obtained according to the area characteristics of the defect area.

2. The method for detecting surface defects of semiconductor ceramic composite materials based on image processing according to claim 1, characterized in that: The step of obtaining the edge intersection point according to the quantity characteristics of other edge pixels within the neighborhood of the edge pixel point on the edge line in the appearance image comprises: If there are no less than three other edge pixels within the eight-neighborhood range of any edge pixel point, the arbitrary edge pixel point is an edge intersection point.

3. The method for detecting surface defects of semiconductor ceramic composite materials based on image processing according to claim 1, characterized in that: The step of acquiring different target edges in the edge line according to the edge intersection points comprises: Taking any edge intersection as a starting point, a local edge line from the arbitrary edge intersection to other directly connected edge intersections is taken as the target edge.

4. The method for detecting surface defects of semiconductor ceramic composite materials based on image processing according to claim 1, characterized in that: The step of obtaining a local gradient magnitude difference value and a local gradient direction difference value according to the gradient difference characteristics of adjacent edge pixels in the target edge comprises: , the gradient amplitude and gradient direction of the edge pixel points in the target edge are respectively used as gradient eigenvalues; where, Represents the gradient eigenvalue of the i-th edge pixel in the target edge, Indicates The gradient eigenvalue of edge pixels, Indicates The gradient eigenvalue of edge pixels; Indicates and The difference of the gradient eigenvalues ​​of edge pixels, represents the i-th and The difference of the gradient eigenvalues ​​of edge pixels, represents the i-th and The difference between the gradient eigenvalues ​​of edge pixels; when the gradient eigenvalue represents the gradient amplitude, Represents the local gradient difference value of the i-th edge pixel. When the gradient eigenvalue represents the gradient direction, Represents the local gradient direction difference value of the i-th edge pixel.

5. The method for detecting surface defects of semiconductor ceramic composite materials based on image processing according to claim 1, characterized in that: The step of obtaining edge smoothing feature values ​​according to the distribution characteristics of local gradient magnitude difference values ​​and local gradient direction difference values ​​of edge pixels in the target edge comprises: The standard deviation of the local gradient magnitude difference values ​​of the edge pixels in the target edge is calculated to obtain a first distribution disorder; the standard deviation of the local gradient direction difference values ​​of the edge pixels in the target edge is calculated to obtain a second distribution disorder; the sum of the first distribution disorder and the second distribution disorder is calculated and negatively correlated to obtain an edge smoothing feature value of the target edge.

6. The method for detecting surface defects of semiconductor ceramic composite materials based on image processing according to claim 1, characterized in that: The step of obtaining the smoothness difference according to the difference characteristics of the edge smoothing feature values ​​of any two target edges comprises: The absolute value of the difference between the edge smoothing feature values ​​of any two target edges is calculated to obtain the smoothing difference between the any two target edges.

7. The method for detecting surface defects of semiconductor ceramic composite materials based on image processing according to claim 1, characterized in that: The step of obtaining a defect probability value according to the discrete characteristics of the smooth difference and judging whether there is a defect comprises: The difference between the maximum value and the minimum value of the smoothing difference is calculated and normalized to obtain a defect probability value of the composite material appearance; when the defect probability value does not exceed a preset normal threshold, there is a defect in the composite material appearance.

8. The method for detecting surface defects of semiconductor ceramic composite materials based on image processing according to claim 1, characterized in that: The steps of clustering the target edges according to the smoothness difference and selecting defect edge clusters from the clustering results according to the edge smoothness feature values ​​include: Clustering is performed using a K-means algorithm according to the smoothing differences between target edges to obtain two target edge clusters; and the target edge cluster containing the minimum value of the edge smoothing eigenvalue is used as a defect edge cluster.

9. The method for detecting surface defects of semiconductor ceramic composite materials based on image processing according to claim 1, characterized in that: The step of obtaining the seed point of the watershed segmentation algorithm according to the grayscale difference characteristics of the areas on both sides of any defect edge in the defect edge cluster comprises: Taking the middle edge pixel point of any defective edge as the starting point, taking a preset number of connected pixel points along the gradient direction of the middle edge pixel point and constructing a pixel point sequence; taking a preset number of connected pixel points along the opposite direction of the gradient direction of the middle edge pixel point and constructing another pixel point sequence; differentiating the grayscale values ​​of the pixel point sequence and calculating the standard deviation to obtain the grayscale chaos degree; taking any pixel point in the pixel point sequence corresponding to the maximum value of the grayscale chaos degree as the seed point.

10. The method for detecting surface defects of semiconductor ceramic composite materials based on image processing according to claim 1, characterized in that: The step of obtaining the degree of defect according to the area characteristics of the defect area comprises: The ratio of the number of pixels in the defect area to the number of pixels in the appearance image is calculated to obtain the degree of the defect.

Images