A plate defect detection system and detection method based on multi-feature fusion

Abstract

The present invention discloses a plate defect detection system and detection method based on multi-feature fusion, the system comprising: an image acquisition and preprocessing module, a chromaticity coordinate and texture feature acquisition module and a defect detection marking module; the image acquisition and preprocessing module acquires the image information of the target plate and preprocesses the image information; the chromaticity coordinate and texture feature acquisition module calculates the preprocessed image information to obtain a chromaticity library and a texture defect model; the defect detection marking module detects defects in the image information and marks the locations of the defects. The present invention establishes a plate defect detection system, establishes a chromaticity library and a texture feature model by image processing, has fast processing speed, high precision and good visual effect.

Description

Technical Field

[0001] The present invention relates to the technical field of defect detection, and more specifically, to a plate defect detection system and detection method based on multi-feature fusion. Background Art

[0002] In recent years, with the continuous improvement of people's living standards, people's demand for wood materials has been growing. In the decoration market, due to the increasing scarcity of wood resources and the fact that wood is prone to various defects during its growth, the supply of high-quality wood is in short supply, resulting in a continuous increase in the price of natural wood. Therefore, in order to improve the comprehensive utilization rate of wood and reduce the price of boards, artificial wood boards are made from wood and other non-wood plants. Artificial wood boards mainly include three categories of products: plywood, particleboard and fiberboard. Their surfaces need to be decorated with imitation wood grain to meet the actual use requirements. Artificial wood boards have been mass-produced in an automated manner, which can easily lead to defects such as defects, scratches, debris and oil stains during the production process. Therefore, defect detection on the surface of artificial wood boards is an important part of wooden finished product inspection.

[0003] At present, the methods of plate defect detection mainly include manual detection, ultrasonic detection, and X-ray detection. However, due to the different types of plates and the different types, shapes, and sizes of defects on the plates, the accuracy and efficiency of manual detection are very low and cannot meet the purpose of real-time detection of the production line. Ultrasonic detection mainly detects internal defects of plates, and there are still technical difficulties to be overcome for surface defect detection. X-ray detection can effectively improve the efficiency and accuracy of detection, but the cost will become very high and it is not suitable for large-scale factory production applications. With the development of electronic components and the application of machine vision in image processing, detection methods based on machine vision have occupied a more important position in industrial quality inspection.

[0004] In order to solve the above problems of time-consuming, labor-intensive, poor results, and the inevitable failure to detect defects in the artificial wood board detection process, a multi-feature fusion board defect detection system can be designed. Based on features such as color, mean, standard value, and smoothness, the system uses a random forest algorithm to combine texture feature information in time and space, and then establishes a texture feature model and color library of the board. The system uses computer automatic processing instead of manual operation, which not only saves time and effort, but also can achieve higher processing accuracy and visual effects. Summary of the invention

[0005] In order to solve at least one of the above technical problems, the present invention proposes a plate defect detection system and detection method based on multi-feature fusion.

[0006] A first aspect of the present invention provides a plate defect detection system based on multi-feature fusion, comprising:

[0007] An image acquisition and preprocessing module is used to acquire image information containing a target plate and preprocess the image information;

[0008] Chromaticity coordinate and texture feature acquisition module, used to generate the chromaticity library and texture defect model of the board;

[0009] The defect detection and marking module is used to detect defects in the image information and mark the defect positions.

[0010] In this solution, the image acquisition and preprocessing module is implemented based on a binocular camera system;

[0011] Obtain the intrinsic parameters and extrinsic parameters of the camera, and obtain the mapping relationship of each point in the image information of the target plate in the coordinate transformation according to the intrinsic parameters and extrinsic parameters of the camera;

[0012] Obtaining distortion-corrected image information through inverse transformation according to the mapping relationship, and extracting a region of interest through the distortion-corrected image information;

[0013] In camera calibration, the left and right cameras in the binocular camera system are unified into the same world coordinate system according to the change of phase, and the relative translation of the two images is calculated by calculating the relative position of the image information obtained by the left and right cameras in the world coordinate system;

[0014] According to the relative translation amount, the regions of interest in the image information collected by the left and right cameras are stitched to generate a complete target plate image.

[0015] In this solution, the chromaticity library is generated by the chromaticity coordinate and texture feature acquisition module, specifically:

[0016] Acquire a plate image data set, import the plate image data set into the chromaticity coordinate and texture feature acquisition module, and calculate the chromaticity coordinate of each pixel in each picture in the LUV color space;

[0017] A chromaticity library is established through the chromaticity coordinates of each image.

[0018] In this solution, the plate texture defect model is generated by the chromaticity coordinate and texture feature acquisition module, specifically:

[0019] Obtain the plate image information containing surface defects in the plate image data set, and calculate the mean, standard value, smoothness, skewness and root mean square value of the surface defect part;

[0020] The mean, standard value, smoothness, skewness and root mean square value are input into a random forest model as training data for training to obtain a texture defect model.

[0021] This plan also includes:

[0022] Multiple subsamples are formed in the training data through the Bootstrap method;

[0023] Randomly extracting d feature elements from the mean, standard value, smoothness, skewness and root mean square value as candidate attributes of the decision tree, calculating the Gini index of the candidate attribute, splitting the decision tree nodes, and repeatedly splitting the decision tree until n decision trees are generated;

[0024] The decision trees are connected to obtain classification results, and the classification results are voted to obtain models of surface scratches, surface oil stains, plate defects and missing patterns.

[0025] In this solution, the defect location is marked by the defect detection marking module, specifically:

[0026] The chromaticity value and chromaticity coordinates of each pixel in the acquired complete target plate image are compared with the chromaticity library in the texture feature acquisition module for the first time, and the texture parameters in the complete target plate image are calculated and compared with the texture defect model in the chromaticity coordinate and texture feature acquisition module for the second time;

[0027] If there is a defect in the first comparison test, the defective part is marked, and a second comparison test is performed to determine whether there is a defect in the second comparison test. If there is a defect, the defective part is marked, and the plate inspection result is generated in combination with the defect mark in the first comparison test. If there is no defect, the plate inspection result is generated according to the defect mark in the first comparison test;

[0028] If there is no defect in the first comparative test, a second comparative test is performed to determine whether there is a defect in the second comparative test. If there is a defect, a plate inspection result is generated based on the defect mark in the second comparative test. If there is no defect, a plate inspection result indicating that the plate has no defects is generated.

[0029] The second aspect of the present invention further provides a plate defect detection method based on multi-feature fusion, which is applied to a plate defect detection system based on multi-feature fusion, and comprises the following steps:

[0030] Acquire a plate image data set, calculate the chromaticity coordinates in each image according to the plate image data set, and establish a chromaticity library;

[0031] The mean, standard value, smoothness, skewness and root mean square value of the surface defect part are calculated and input into the random forest model, and the texture defect model is obtained through calculation and analysis;

[0032] The left and right cameras in the binocular camera system respectively collect image information containing the target plate, and perform distortion correction on the collected image information to obtain a corrected image;

[0033] Extract the region of interest from the corrected image information, and stitch the regions of interest in the image information collected by the left and right cameras to obtain a complete target plate image;

[0034] Compare and inspect the complete target plate image with the colorimetric library and texture defect model, mark the defects based on the comparison test results and generate the plate inspection results;

[0035] Determine whether the camera acquisition signal continues to be transmitted in the PLC. If so, continue to perform defect detection on the image information of the plate. If not, output the plate detection result and end the detection.

[0036] In this solution, the establishment of the chromaticity library is specifically as follows:

[0037] Acquire a plate image data set, import the plate image data set into the chromaticity coordinate and texture feature acquisition module, and calculate the chromaticity coordinate of each pixel in each picture in the LUV color space;

[0038] A chromaticity library is established through the chromaticity coordinates of each image.

[0039] In this solution, the texture defect model is specifically:

[0040] Obtain the plate image information containing surface defects in the plate image data set, and calculate the mean, standard value, smoothness, skewness and root mean square value of the surface defect part;

[0041] The calculated mean, standard value, smoothness, skewness and RMS value are used as input to the random forest;

[0042] Multiple subsamples are formed in the training data through the Bootstrap method;

[0043] Randomly extracting d feature elements from the mean, standard value, smoothness, skewness and root mean square value as candidate attributes of the decision tree, calculating the Gini index of the candidate attribute, splitting the decision tree nodes, and repeatedly splitting the decision tree until n decision trees are generated;

[0044] The decision trees are connected to obtain classification results, and the classification results are voted to obtain models of surface scratches, surface oil stains, plate defects and missing patterns.

[0045] In this solution, the complete target plate image is compared with the color library and the texture defect model, and the defects are marked according to the comparison test results and the plate test results are generated, specifically:

[0046] The chromaticity value and chromaticity coordinates of each pixel in the acquired complete target plate image are compared with the chromaticity library in the texture feature acquisition module for the first time, and the texture parameters in the complete target plate image are calculated and compared with the texture defect model in the chromaticity coordinate and texture feature acquisition module for the second time;

[0047] If there is a defect in the first comparison test, the defective part is marked, and a second comparison test is performed to determine whether there is a defect in the second comparison test. If there is a defect, the defective part is marked, and the plate inspection result is generated in combination with the defect mark in the first comparison test. If there is no defect, the plate inspection result is generated according to the defect mark in the first comparison test;

[0048] If there is no defect in the first comparative test, a second comparative test is performed to determine whether there is a defect in the second comparative test. If there is a defect, a plate inspection result is generated based on the defect mark in the second comparative test. If there is no defect, a plate inspection result indicating that the plate has no defects is generated.

[0049] The present invention establishes a chromaticity library by calculating the chromaticity coordinates of the board, extracts texture parameters of the board surface from multiple angles, and uses a random forest model to combine texture feature information in time and space to obtain more complete texture features, thereby establishing a texture feature model of the board and realizing the automation of defect detection of artificial wood boards. The board defect detection system of the present invention has fast processing speed, high accuracy and good visual effect; it replaces tedious, time-consuming and laborious manual operations, and can detect whether there are defects on the board surface in batches, which plays a positive role in promoting its further development. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 A schematic diagram of a plate defect detection system based on multi-feature fusion according to the present invention is shown;

[0051] Figure 2 The flowchart of establishing the texture feature model in the present invention is shown;

[0052] Figure 3 A flow chart of a plate defect detection method based on multi-feature fusion of the present invention is shown. DETAILED DESCRIPTION

[0053] In order to more clearly understand the above-mentioned purpose, features and advantages of the present invention, the present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be noted that the embodiments of the present application and the features in the embodiments can be combined with each other without conflict.

[0054] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Therefore, the protection scope of the present invention is not limited to the specific embodiments disclosed below.

[0055] The terms "first", "second", "third", etc. may be used herein to describe various elements, components and / or layers, which should not be limited by these terms. These terms are used to distinguish one element, component or layer from another element, component or layer. Therefore, the first element, component or layer described below may be referred to as a second element, component or layer without departing from the scope of the present disclosure.

[0056] Figure 1 A schematic diagram of a plate defect detection system based on multi-feature fusion according to the present invention is shown.

[0057] like Figure 1 The first aspect of the present invention provides a plate defect detection system based on multi-feature fusion, comprising the following modules:

[0058] An image acquisition and preprocessing module is used to acquire image information containing a target plate and preprocess the image information;

[0059] Chromaticity coordinate and texture feature acquisition module, used to generate the chromaticity library and texture defect model of the plate;

[0060] The defect detection and marking module is used to detect defects in the image information and mark the defect positions.

[0061] It should be noted that the image acquisition and preprocessing module is implemented based on a binocular camera system; the camera's internal and external parameters are obtained. The internal parameters of each camera are unique, including three radial distortion coefficients and two tangential distortion parameters, and the external parameters are not fixed, including rotation vectors and translation vectors. The image can be calculated from the camera coordinate system through the internal and external parameters of the camera, and the mapping relationship of each point in the image information of the target plate in the coordinate transformation is obtained according to the internal and external parameters of the camera; the image information after distortion correction is obtained through inverse transformation according to the mapping relationship, and the region of interest is extracted through the image information after distortion correction; in camera calibration, the left and right cameras in the binocular camera system are unified to the same world coordinate system according to the change of phase, and the relative translation of the two images is calculated by calculating the relative position of the image information obtained by the left and right cameras in the world coordinate system; according to the relative translation, the region of interest in the image information collected by the left and right cameras is stitched to generate a complete target plate image.

[0062] It should be noted that the chromaticity library is generated by the chromaticity coordinate and texture feature acquisition module. The chromaticity library is a database formed by the chromaticity coordinates of each pixel point, specifically:

[0063] A plate image data set is obtained, and the plate image data set is imported into the chromaticity coordinate and texture feature acquisition module to calculate the chromaticity coordinate of each pixel in each picture in the LUV color space; and a chromaticity library is established through the chromaticity coordinates of each image.

[0064] Figure 2 The flowchart of establishing the texture feature model in the present invention is shown.

[0065] It should be noted that the plate texture defect model is generated by the chromaticity coordinate and texture feature acquisition module, specifically:

[0066] The plate image information containing surface defects in the plate image data set is obtained, and the mean, standard value, smoothness, skewness and root mean square value of the surface defect part are calculated; the mean, standard value, smoothness, skewness and root mean square value are input into the random forest model as training data for training, so as to obtain a texture defect model.

[0067] A plurality of subsamples are formed in the training data by using the Bootstrap method; d feature elements are randomly selected from the mean, standard value, smoothness, skewness and root mean square value as candidate attributes of the decision tree, the Gini index of the candidate attribute is calculated, the decision tree nodes are split, and the decision tree is split repeatedly until n decision trees are generated; the decision trees are connected to obtain classification results, and the classification results are voted to obtain models of surface scratches, surface oil stains, plate defects and missing patterns.

[0068] It should be noted that the defect location is marked by the defect detection marking module, specifically:

[0069] The chromaticity value and chromaticity coordinates of each pixel in the acquired complete target plate image are compared with the chromaticity library in the texture feature acquisition module for the first time, and the texture parameters in the complete target plate image are calculated and compared with the texture defect model in the chromaticity coordinate and texture feature acquisition module for the second time;

[0070] If there is a defect in the first comparison test, the defective part is marked, and a second comparison test is performed to determine whether there is a defect in the second comparison test. If there is a defect, the defective part is marked, and the plate inspection result is generated in combination with the defect mark in the first comparison test. If there is no defect, the plate inspection result is generated according to the defect mark in the first comparison test;

[0071] If there is no defect in the first comparative test, a second comparative test is performed to determine whether there is a defect in the second comparative test. If there is a defect, a plate inspection result is generated based on the defect mark in the second comparative test. If there is no defect, a plate inspection result indicating that the plate has no defects is generated.

[0072] Figure 3 A flow chart of a plate defect detection method based on multi-feature fusion of the present invention is shown.

[0073] The second aspect of the present invention further provides a plate defect detection method based on multi-feature fusion, which is applied to a plate defect detection system based on multi-feature fusion, and comprises the following steps:

[0074] Acquire a plate image data set, calculate the chromaticity coordinates in each image according to the plate image data set, and establish a chromaticity library;

[0075] The mean, standard value, smoothness, skewness and root mean square value of the surface defect part are calculated and input into the random forest model, and the texture defect model is obtained through calculation and analysis;

[0076] The left and right cameras in the binocular camera system respectively collect image information containing the target plate, and perform distortion correction on the collected image information to obtain a corrected image;

[0077] Extract the region of interest from the corrected image information, and stitch the regions of interest in the image information collected by the left and right cameras to obtain a complete target plate image;

[0078] Compare and inspect the complete target plate image with the colorimetric library and texture defect model, mark the defects based on the comparison test results and generate the plate inspection results;

[0079] Determine whether the camera acquisition signal continues to be transmitted in the PLC. If so, continue to perform defect detection on the image information of the plate. If not, output the plate detection result and end the detection.

[0080] It should be noted that the establishment of the chromaticity library is specifically as follows:

[0081] A plate image data set is obtained, and the plate image data set is imported into the chromaticity coordinate and texture feature acquisition module to calculate the chromaticity coordinate of each pixel in each picture in the LUV color space; and a chromaticity library is established through the chromaticity coordinates of each image.

[0082] It should be noted that the texture defect model is specifically:

[0083] Obtain the plate image information containing surface defects in the plate image data set, and calculate the mean, standard value, smoothness, skewness and root mean square value of the surface defect part;

[0084] The calculated mean, standard value, smoothness, skewness and RMS value are used as input to the random forest;

[0085] Multiple subsamples are formed in the training data through the Bootstrap method;

[0086] Randomly extracting d feature elements from the mean, standard value, smoothness, skewness and root mean square value as candidate attributes of the decision tree, calculating the Gini index of the candidate attribute, splitting the decision tree nodes, and repeatedly splitting the decision tree until n decision trees are generated;

[0087] The decision trees are connected to obtain classification results, and the classification results are voted to obtain models of surface scratches, surface oil stains, plate defects and missing patterns.

[0088] It should be noted that the complete target plate image is compared with the color library and the texture defect model, and the defects are marked according to the comparison test results and the plate test results are generated, specifically:

[0089] The chromaticity value and chromaticity coordinates of each pixel in the acquired complete target plate image are compared with the chromaticity library in the texture feature acquisition module for the first time, and the texture parameters in the complete target plate image are calculated and compared with the texture defect model in the chromaticity coordinate and texture feature acquisition module for the second time;

[0090] If there is a defect in the first comparison test, the defective part is marked, and a second comparison test is performed to determine whether there is a defect in the second comparison test. If there is a defect, the defective part is marked, and the plate inspection result is generated in combination with the defect mark in the first comparison test. If there is no defect, the plate inspection result is generated according to the defect mark in the first comparison test;

[0091] If there is no defect in the first comparative test, a second comparative test is performed to determine whether there is a defect in the second comparative test. If there is a defect, a plate inspection result is generated based on the defect mark in the second comparative test. If there is no defect, a plate inspection result indicating that the plate has no defects is generated.

[0092] In the several embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as: multiple units or components can be combined, or can be integrated into another system, or some features can be ignored or not executed. In addition, the coupling, direct coupling, or communication connection between the components shown or discussed can be through some interfaces, and the indirect coupling or communication connection of the devices or units can be electrical, mechanical or other forms.

[0093] The units described above as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units; they may be located in one place or distributed on multiple network units; some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.

[0094] In addition, all functional units in the embodiments of the present invention may be integrated into one processing unit, or each unit may be separately used as a unit, or two or more units may be integrated into one unit; the above-mentioned integrated units may be implemented in the form of hardware or in the form of hardware plus software functional units.

[0095] Those skilled in the art can understand that: all or part of the steps of implementing the above method embodiments can be completed by hardware related to program instructions, and the aforementioned program can be stored in a computer-readable storage medium. When the program is executed, it executes the steps of the above method embodiments; and the aforementioned storage medium includes: mobile storage devices, read-only memories (ROM, Read-Only Memory), random access memories (RAM, Random Access Memory), disks or optical disks, and other media that can store program codes.

[0096] Alternatively, if the above-mentioned integrated unit of the present invention is implemented in the form of a software function module and sold or used as an independent product, it can also be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the embodiment of the present invention can be essentially or partly reflected in the form of a software product that contributes to the prior art. The computer software product is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the methods described in each embodiment of the present invention. The aforementioned storage medium includes: various media that can store program codes, such as mobile storage devices, ROM, RAM, magnetic disks or optical disks.

[0097] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims

1. A plate defect detection system based on multi-feature fusion, characterized in that: include: An image acquisition and preprocessing module is used to acquire image information containing a target plate and preprocess the image information; Chromaticity coordinate and texture feature acquisition module, used to generate the chromaticity library and texture defect model of the board; A defect detection and marking module, used to detect defects in the image information and mark the defect locations; The image acquisition and preprocessing module is implemented based on a binocular camera system; Obtain the intrinsic parameters and extrinsic parameters of the camera, and obtain the mapping relationship of each point in the image information of the target plate in the coordinate transformation according to the intrinsic parameters and extrinsic parameters of the camera; Obtaining distortion-corrected image information through inverse transformation according to the mapping relationship, and extracting a region of interest through the distortion-corrected image information; In camera calibration, the left and right cameras in the binocular camera system are unified into the same world coordinate system according to the change of phase, and the relative translation of the two images is calculated by calculating the relative position of the image information obtained by the left and right cameras in the world coordinate system; According to the relative translation amount, the regions of interest in the image information collected by the left and right cameras are stitched together to generate a complete target plate image; The plate texture defect model is generated by the chromaticity coordinate and texture feature acquisition module, specifically: Obtain the plate image information containing surface defects in the plate image data set, and calculate the mean, standard value, smoothness, skewness and root mean square value of the surface defect part; The mean, standard value, smoothness, skewness and root mean square value are input into a random forest model as training data for training to obtain a texture defect model; The training of the texture defect model is specifically as follows: Multiple subsamples are formed in the training data through the Bootstrap method; Randomly extracting d feature elements from the mean, standard value, smoothness, skewness and root mean square value as candidate attributes of the decision tree, calculating the Gini index of the candidate attribute, splitting the decision tree nodes, and repeatedly splitting the decision tree until n decision trees are generated; The decision trees are connected to obtain classification results, and the classification results are voted to obtain models of surface scratches, surface oil stains, plate defects and missing patterns.

2. A plate defect detection system based on multi-feature fusion according to claim 1, characterized in that: The chromaticity library is generated by the chromaticity coordinate and texture feature acquisition module, specifically: Acquire a plate image data set, import the plate image data set into the chromaticity coordinate and texture feature acquisition module, and calculate the chromaticity coordinate of each pixel in each picture in the LUV color space; A chromaticity library is established through the chromaticity coordinates of each picture.

3. The plate defect detection system based on multi-feature fusion according to claim 1 is characterized in that: The defect detection and marking module is used to mark the location of the defect, specifically: The chromaticity value and chromaticity coordinates of each pixel in the acquired complete target plate image are compared with the chromaticity library in the texture feature acquisition module for the first time, and the texture parameters in the complete target plate image are calculated and compared with the texture defect model in the chromaticity coordinate and texture feature acquisition module for the second time; If there is a defect in the first comparison test, the defective part is marked, and a second comparison test is performed to determine whether there is a defect in the second comparison test. If there is a defect, the defective part is marked, and the plate inspection result is generated in combination with the defect mark in the first comparison test. If there is no defect, the plate inspection result is generated according to the defect mark in the first comparison test; If there is no defect in the first comparative test, a second comparative test is performed to determine whether there is a defect in the second comparative test. If there is a defect, a plate inspection result is generated based on the defect mark in the second comparative test. If there is no defect, a plate inspection result indicating that the plate has no defects is generated.

4. A plate defect detection method based on multi-feature fusion, characterized in that: A plate defect detection system based on multi-feature fusion as described in any one of claims 1 to 3 above comprises the following steps: Acquire a plate image data set, calculate the chromaticity coordinates in each picture according to the plate image data set, and establish a chromaticity library; The mean, standard value, smoothness, skewness and root mean square value of the surface defect part are calculated and input into the random forest model, and the texture defect model is obtained through calculation and analysis; The left and right cameras in the binocular camera system respectively collect image information containing the target plate, and perform distortion correction on the collected image information to obtain a corrected image; Extract the region of interest from the corrected image information, and stitch the regions of interest in the image information collected by the left and right cameras to obtain a complete target plate image; Compare and inspect the complete target plate image with the colorimetric library and texture defect model, mark the defects based on the comparison test results and generate the plate inspection results; Determine whether the camera acquisition signal continues to be transmitted in the PLC. If so, continue to perform defect detection on the image information of the plate. If not, output the plate detection result and end the detection.

5. The plate defect detection method based on multi-feature fusion according to claim 4 is characterized in that: The establishment of the chromaticity library is specifically as follows: Acquire a plate image data set, import the plate image data set into the chromaticity coordinate and texture feature acquisition module, and calculate the chromaticity coordinate of each pixel in each picture in the LUV color space; A chromaticity library is established through the chromaticity coordinates of each picture.

6. The plate defect detection method based on multi-feature fusion according to claim 4 is characterized in that: The texture defect model is specifically: Obtain the plate image information containing surface defects in the plate image data set, and calculate the mean, standard value, smoothness, skewness and root mean square value of the surface defect part; The calculated mean, standard value, smoothness, skewness and RMS value are used as input to the random forest; Multiple subsamples are formed in the training data through the Bootstrap method; Randomly extracting d feature elements from the mean, standard value, smoothness, skewness and root mean square value as candidate attributes of the decision tree, calculating the Gini index of the candidate attribute, splitting the decision tree nodes, and repeatedly splitting the decision tree until n decision trees are generated; The decision trees are connected to obtain classification results, and the classification results are voted to obtain models of surface scratches, surface oil stains, plate defects and missing patterns.

7. The plate defect detection method based on multi-feature fusion according to claim 4 is characterized in that: The complete target plate image is compared with the colorimetric library and the texture defect model, and the defects are marked according to the comparison test results and the plate test results are generated, specifically: The chromaticity value and chromaticity coordinates of each pixel in the acquired complete target plate image are compared with the chromaticity library in the texture feature acquisition module for the first time, and the texture parameters in the complete target plate image are calculated and compared with the texture defect model in the chromaticity coordinate and texture feature acquisition module for the second time; If there is a defect in the first comparison test, the defective part is marked, and a second comparison test is performed to determine whether there is a defect in the second comparison test. If there is a defect, the defective part is marked, and the plate inspection result is generated in combination with the defect mark in the first comparison test. If there is no defect, the plate inspection result is generated according to the defect mark in the first comparison test; If there is no defect in the first comparative test, a second comparative test is performed to determine whether there is a defect in the second comparative test. If there is a defect, a plate inspection result is generated based on the defect mark in the second comparative test. If there is no defect, a plate inspection result indicating that the plate has no defects is generated.

Images