Search

CN-121998981-A - Composite board ceramic coating quality detection method and system

CN121998981ACN 121998981 ACN121998981 ACN 121998981ACN-121998981-A

Abstract

The invention belongs to the technical field of coating quality detection, and particularly relates to a composite board ceramic coating quality detection method and system, wherein the method comprises the steps of obtaining local neighborhood information of each pixel point according to a sliding window, calculating gray local contrast and constructing gray suppression factors; the method comprises the steps of taking a normalized average value of pixel gradient amplitude values in a sliding window as a texture feature index, obtaining a background consistency enhancement coefficient based on gray scale inhibition factors of normal and abnormal candidate sets and standard deviations of all gray scale inhibition factors, obtaining a defect texture protection coefficient by combining the texture feature index of a missing coating candidate set and standard deviations of all texture feature indexes, constructing sparse punishment weight of each pixel, weighting regularization parameters to obtain dynamic regularization parameters, constructing and solving an objective function to obtain a sparse abnormal matrix, and separating a missing coating area from the sparse abnormal matrix by combining the texture feature index. The invention improves the accuracy of the quality detection of the composite board ceramic coating.

Inventors

  • HAO BAOGUO
  • GAO HAIPENG
  • HAO NING
  • LEI YUNLONG

Assignees

  • 宝鸡市力合金属复合材料股份有限公司

Dates

Publication Date
20260508
Application Date
20260409

Claims (10)

  1. 1. The composite board ceramic coating quality detection method is characterized by comprising the following steps of: The method comprises the steps of obtaining a gray analysis image of a ceramic coating, defining a sliding window on the gray analysis image, extracting local neighborhood information of each pixel point, obtaining gray local contrast of each pixel point based on the local neighborhood information and gray values of the pixel points, constructing a gray suppression factor of each pixel point according to the distribution condition of the gray local contrast of all the pixel points, and taking a normalization result of the average value of gradient amplitude values of all the pixel points in the sliding window as a texture characteristic index of each pixel point; Obtaining a defect texture protection coefficient according to the difference of texture characteristic indexes of the missing candidate set and the reflective candidate set and the standard deviation of texture characteristic indexes of all pixel points, and constructing sparse punishment weight of each pixel point by combining the background consistency enhancement coefficient, the gray suppression factor and the texture characteristic indexes; the method comprises the steps of weighting globally fixed regularization parameters through sparse punishment weights to obtain dynamic regularization parameters, constructing an optimized objective function according to the dynamic regularization parameters, solving the objective function to obtain a sparse abnormal matrix, separating a coating missing region from the sparse abnormal matrix by combining texture characteristic indexes, and outputting the coating missing region as a composite board ceramic coating quality detection result.
  2. 2. The method for detecting the quality of the ceramic coating of the composite board according to claim 1, wherein the method for obtaining the gray local contrast of each pixel point comprises the following steps: ; Wherein, the Is a pixel point Gray local contrast of (2); Is a pixel point Gray values of (2); Is a peripheral pixel point Gray values of (2); Is an index of a pixel point located at the center of the sliding window; is the index of the peripheral pixel points except the center of the sliding window; Is that Sliding window of size, wherein An odd number greater than 1; Is a pixel point And peripheral pixel points Euclidean distance between them; Is formed by pixel points Sliding window with center In, pixel point Standard deviation of euclidean distance from all peripheral pixel points; Is a sliding window The pixel point is included The total number of all pixels within; Is a natural exponential function; Is an absolute value function.
  3. 3. The method for detecting the quality of the ceramic coating of the composite board according to claim 1, wherein the construction of the gray scale suppression factor of each pixel according to the distribution condition of the gray scale local contrast of all the pixels comprises the following steps: ; Wherein, the Is a pixel point Gray scale inhibitor of (c); Is a pixel point Gray local contrast of (2); Is the average value of gray local contrast of all pixel points; is the standard deviation of gray local contrast of all pixel points; Is a natural exponential function.
  4. 4. The method for detecting the quality of the ceramic coating of the composite board according to claim 1, wherein the method for acquiring the normal candidate set and the abnormal candidate set comprises the following steps: Calculating the average value and standard deviation of gray scale inhibition factors of all pixel points, respectively recording as Setting a normal threshold according to the material characteristics and the detection precision requirements of the composite board ceramic coating The normal threshold value The value range of (2) is set as To the point of The pixels with the gray level inhibition factors larger than or equal to the normal threshold value are divided into normal candidate sets belonging to the normal coating areas, and the pixels with the gray level inhibition factors smaller than the normal threshold value are divided into abnormal candidate sets belonging to the abnormal areas.
  5. 5. The method for detecting the quality of the ceramic coating of the composite board according to claim 1, wherein the obtaining the background consistency enhancement coefficient according to the difference of the gray scale inhibitor of the normal candidate set and the gray scale inhibitor of the abnormal candidate set and the standard deviation of the gray scale inhibitor of all the pixel points comprises the following steps: ; Wherein, the Is a background consistency enhancement factor; is the average value of all gray scale inhibition factors in the normal candidate set; is the average value of all gray scale inhibition factors in the abnormal candidate set; is the standard deviation of the gray scale inhibitor for all pixels.
  6. 6. The method for detecting the quality of the ceramic coating of the composite board according to claim 1, wherein the method for acquiring the missing coating candidate set and the reflective candidate set comprises the following steps: Obtaining standard deviation of texture characteristic indexes of all pixel points, and marking the standard deviation as Then, aiming at an abnormal candidate set belonging to an abnormal region, obtaining a texture feature index of each pixel point in the abnormal candidate set, and calculating an abnormal threshold value by using an Ojin threshold algorithm on the distribution of the texture feature index And making the texture characteristic index larger than or equal to the abnormality threshold value Dividing the pixel points of the pattern into a missing coating candidate set belonging to a missing coating region, and setting the texture characteristic index smaller than an abnormal threshold value Is divided into a reflection candidate set belonging to a reflection area.
  7. 7. The method for detecting the quality of the ceramic coating of the composite board according to claim 1, wherein the obtaining the defect texture protection coefficient according to the difference of the texture characteristic indexes of the missing coating candidate set and the reflective candidate set and the standard deviation of the texture characteristic indexes of all pixel points comprises the following steps: ; Wherein, the Is a defect texture protection factor; is the average value of texture characteristic indexes of all pixel points of the missing coating candidate set; Is the average value of texture characteristic indexes of all pixel points in the reflection candidate set; Is the standard deviation of the texture characteristic index of all the pixel points.
  8. 8. The method for detecting the quality of the ceramic coating of the composite board according to claim 1, wherein the constructing the sparse penalty weight of each pixel point comprises the following steps: ; Wherein, the Is a pixel point Is a sparse penalty weight of (1); Is a pixel point Gray scale inhibitor of (c); Is a pixel point Texture feature index of (2); is a background consistency enhancement factor; Is a defect texture protection factor; Is a natural exponential function.
  9. 9. The method for detecting the quality of the ceramic coating of the composite board according to claim 1, wherein the step of separating the missing coating area from the sparse anomaly matrix by combining the texture characteristic indexes comprises the following steps: Calculating the average value and standard deviation of texture characteristic indexes of all pixel points, setting a double-adaptive threshold value, namely, a missing coating judging threshold value is the average value plus 1.5 times of standard deviation, a reflection judging threshold value is the average value minus 1.5 times of standard deviation, the 1.5 times of standard deviation can be adjusted according to actual conditions, matching corresponding texture characteristic index values of abnormal pixel points marked as 1 in a sparse abnormal matrix, screening the abnormal pixel points, namely, keeping the mark of the texture characteristic index in the sparse abnormal matrix 1 when the texture characteristic index reaches or exceeds the missing coating judging threshold value, setting the marked 1 of the texture characteristic index to be 0 when the texture characteristic index is lower than or equal to the reflection judging threshold value, removing reflection interference, and after screening, only keeping the mark of the sparse abnormal matrix corresponding to a missing coating area, and further separating the missing coating area from the sparse abnormal matrix.
  10. 10. A composite board ceramic coating quality inspection system comprising a processor and a memory, the memory storing computer program instructions that when executed by the processor implement a composite board ceramic coating quality inspection method according to any one of claims 1-9.

Description

Composite board ceramic coating quality detection method and system Technical Field The invention relates to the technical field of coating quality detection, in particular to a composite board ceramic coating quality detection method and system. Background In the production process, the ceramic coating is neglected to be coated, the metal substrate is exposed to be a very serious quality defect, the quality problems of scratches, corrosion and the like of the composite board in a time far below the design life are easily caused, and safety accidents are seriously caused. Because the metal substrate produces strong light spots in the coating missing area when slightly dithered, the detection accuracy of the quality of the ceramic coating on the surface of the composite board is interfered, for this purpose, the prior art adopts a Robust Principal Component Analysis (RPCA) algorithm to decompose the surface image of the composite board into a low-rank background matrix representing a normal coating and a sparse front Jing Juzhen representing a coating defect, so as to realize accurate separation. However, since conventional RPCA algorithms typically employ globally fixed regularization parametersThe method comprises the steps of balancing weights of a low-rank term and a sparse term, wherein the data characteristics of strong reflection noise points with higher surface brightness and larger area of a metal substrate and the data characteristics of missing coating defects with lower brightness and smaller area are huge, if a larger regularization parameter is set to inhibit strong reflection, an algorithm can filter tiny real missing coating as a background to cause the missing detection rate to rise, if a smaller regularization parameter is set to retain tiny details, the algorithm can misjudge metal reflection light spots as sparse defects to cause the false detection rate to rise, and the accuracy of detecting the quality of ceramic coating on the surface of a composite board can be reduced. Disclosure of Invention The invention provides a composite board ceramic coating quality detection method, which aims to solve the technical problems that the conventional RPCA algorithm is difficult to balance between inhibiting strong reflection noise points and retaining tiny real defects due to the use of fixed regularization parameters, so that the omission rate or false detection rate is increased, and finally the quality detection accuracy of a composite board ceramic coating is affected. The invention provides a composite board ceramic coating quality detection method which comprises the steps of obtaining a gray analysis image, defining a sliding window on the gray analysis image, extracting local neighborhood information of each pixel point, obtaining gray local contrast of each pixel point based on the local neighborhood information and gray values of the pixel points, constructing a gray suppression factor of each pixel point according to distribution conditions of the gray local contrast of all the pixel points, taking a normalized result of an average value of gradient amplitude values of all the pixel points in the sliding window as a texture feature index of each pixel point, obtaining a background consistency enhancement coefficient according to differences of gray suppression factors of a normal candidate set and an abnormal candidate set and standard deviation of the gray suppression factors of all the pixel points, obtaining a defect texture protection coefficient according to differences of the texture feature indexes of a missing candidate set and a reflective candidate set and standard deviation of the texture feature indexes of all the pixel points, combining the background consistency enhancement coefficient, the gray suppression factors and the texture feature indexes, constructing a sparse penalty weight of each pixel point, carrying out sparse regularization on a global fixed weight, carrying out regularization on a regularized matrix according to the sparse weight, obtaining a sparse regularized matrix according to the regularized feature matrix, and obtaining a sparse target-feature matrix, and carrying out sparse-based on the sparse-feature matrix, and obtaining a sparse target-feature-based on the sparse target-matrix, and carrying out sparse-feature-based on the target-feature-map. According to the method, pixel local neighborhood information is extracted through a sliding window, gray level inhibition factors are built by combining gray level local contrast, texture characteristic indexes are built based on Sobel operator gradient amplitude, normal coating, reflection and missing coating areas are accurately distinguished and are enhanced, fixed regularization parameters of robust principal component analysis are optimized to dynamic regularization parameters by building differentiated sparse punishment weights, the problem that reflection inhibition and missing coating retention are d