A plastic product micro defect detection device based on super-resolution imaging

Abstract

The present application provides a plastic product micro defect detection device based on super-resolution imaging, which relates to the field of plastic product detection technology. The device includes: a feature acquisition module for generating initial detection configuration parameters; an imaging module for establishing initial imaging through a microscopic imaging unit; a control module for calling a self-attention control unit to perform initial imaging detection; an adaptive reconstruction module for reconstructing detection configuration parameters based on the self-attention control results, re-performing super-resolution imaging, and generating imaging results; a defect detection module for performing defect detection. The present application can solve the technical problem that the poor quality of super-resolution imaging of surface defect features and the poor precision and accuracy of micro defect feature recognition due to the low adaptability of the device detection parameters to the surface defect features of plastic products can be solved. The super-resolution imaging quality of defect features can be improved, so as to achieve the effect of improving the precision and quality of micro defect detection.

Description

Technical Field

[0001] The present application relates to the technical field of plastic product detection, and in particular to a plastic product micro defect detection device based on super-resolution imaging. Background Art

[0002] Super-resolution imaging is a technology that achieves super-resolution imaging by breaking through the optical diffraction limit. It can restore high-quality images from low-resolution images, thereby obtaining clear and detailed images. Therefore, compared with traditional defect detection methods, micro-defect detection of plastic products based on super-resolution imaging can significantly improve the resolution and clarity of images, so that micro-defects on the surface of plastic products can be more accurately captured and identified, thereby more effectively detecting tiny defects.

[0003] At present, the existing plastic product micro-defect detection equipment based on super-resolution imaging has the technical problem that the compatibility between the equipment detection parameters and the surface defect characteristics of the plastic products is low, resulting in poor super-resolution imaging quality of the surface defect characteristics and poor precision and accuracy in identifying the surface micro-defect characteristics. Summary of the invention

[0004] The purpose of this application is to provide a plastic product micro-defect detection device based on super-resolution imaging, so as to solve the technical problems that the existing plastic product micro-defect detection device based on super-resolution imaging has a low degree of adaptability between the equipment detection parameters and the surface defect characteristics of the plastic products, resulting in poor super-resolution imaging quality of the surface defect characteristics, and poor precision and accuracy in identifying the surface micro-defect characteristics.

[0005] In view of the above problems, the present application provides a plastic product micro-defect detection device based on super-resolution imaging.

[0006] In the first aspect, the present application also provides a plastic product micro defect detection device based on super-resolution imaging, which is used to execute a plastic product micro defect detection method based on super-resolution imaging as described in the first aspect, wherein the device includes a structured light illumination unit, a microscopic imaging unit, and a self-attention control unit, including: a feature acquisition module, which is used to acquire the features of the surface to be detected of the plastic product, and generate initial detection configuration parameters based on the features of the surface to be detected; a response module, which is used to control the structured light illumination unit to irradiate the light source of the surface to be detected with the initial detection configuration parameters after the positioning of the plastic product is completed; an imaging module, which is used to receive the light source through the microscopic imaging unit to establish initial imaging; a control module, which is used to call the self-attention control unit, perform imaging detection of the initial imaging, and generate a self-attention control result, wherein the self-attention control unit is initialized based on the features of the surface to be detected before detection; an adaptive reconstruction module, which is used to reconstruct the detection configuration parameters based on the self-attention control results, and control the structured light illumination unit and the microscopic imaging unit to re-perform super-resolution imaging according to the reconstructed detection configuration parameters to generate imaging results; a defect detection module, which is used to receive the imaging results, perform defect detection, and report defect abnormalities according to the defect detection results.

[0007] In a second aspect, the present application provides a method for detecting microscopic defects of plastic products based on super-resolution imaging, wherein the method is implemented by a device for detecting microscopic defects of plastic products based on super-resolution imaging, wherein the device includes a structured light illumination unit, a microscopic imaging unit, and a self-attention control unit, wherein the method includes: obtaining the features of the surface to be detected of the plastic product, and generating initial detection configuration parameters based on the features of the surface to be detected; after the positioning of the plastic product is completed, controlling the structured light illumination unit with the initial detection configuration parameters to irradiate the light source of the surface to be detected; receiving the light source through the microscopic imaging unit to establish initial imaging; calling the self-attention control unit to perform imaging detection of the initial imaging and generate a self-attention control result, wherein the self-attention control unit is initialized based on the features of the surface to be detected before detection; reconstructing the detection configuration parameters based on the self-attention control results, and controlling the structured light illumination unit and the microscopic imaging unit to re-perform super-resolution imaging according to the reconstructed detection configuration parameters to generate imaging results; receiving the imaging results, performing defect detection, and reporting defect abnormalities according to the defect detection results.

[0008] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0009] By acquiring the features of the surface to be detected of the plastic product, initial detection configuration parameters are generated based on the features of the surface to be detected; when the positioning of the plastic product is completed, the structured light illumination unit is controlled by the initial detection configuration parameters to irradiate the light source of the surface to be detected; the light source is received by the microscopic imaging unit to establish initial imaging; the self-attention regulation unit is called to perform imaging detection of the initial imaging to generate a self-attention regulation result, wherein the self-attention regulation unit is initialized based on the features of the surface to be detected before detection; the detection configuration parameters are reconstructed based on the self-attention regulation result, and the structured light illumination unit and the microscopic imaging unit are controlled to re-perform super-resolution imaging according to the reconstructed detection configuration parameters to generate an imaging result; the imaging result is received, defect detection is performed, and defect abnormalities are reported according to the defect detection result. That is to say, by collecting the features of the surface to be inspected of the plastic product, the initial inspection parameters are configured based on the features of the surface to be inspected; then the structured light illumination unit and the microscopic imaging unit are called to perform imaging according to the initial inspection parameters to obtain the initial imaging result; then the initial imaging result is inspected for imaging quality by the self-attention control unit to obtain the self-attention control result; further, the inspection parameters are optimized and adjusted based on the self-attention control result, and the inspection configuration parameters are reconstructed to improve the adaptability of the equipment inspection parameters to the surface defect features of the plastic product, and the inspection surface is re-imaged with super-resolution based on the reconstructed inspection configuration parameters to generate the imaging result; finally, the imaging result is inspected for defects to determine the defect abnormality results. The accuracy of the equipment inspection parameter setting can be improved, thereby improving the super-resolution imaging quality of the surface defect features of the plastic product, achieving the technical goal of accurately identifying the surface defect features, and achieving the technical effect of improving the detection accuracy and quality of microscopic defects of plastic products.

[0010] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented according to the contents of the specification, and in order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are specifically cited below. It should be understood that the content described in this section is not intended to identify the key or important features of the embodiments of the present application, nor is it intended to limit the scope of the present application. Other features of the present application will become easy to understand through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] In order to more clearly illustrate the technical solutions in the present application or the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings in the following description are only exemplary, and for ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying any creative work.

[0012] Figure 1This is a schematic diagram of the structure of a plastic product micro defect detection device based on super-resolution imaging in this application;

[0013] Figure 2 A schematic diagram of a process for detecting microscopic defects of plastic products based on super-resolution imaging in this application;

[0014] Description of reference numerals:

[0015] Feature acquisition module 11, response module 12, imaging module 13, control module 14, adaptive reconstruction module 15, defect detection module 16. DETAILED DESCRIPTION

[0016] This application provides a plastic product micro defect detection device based on super-resolution imaging, which solves the technical problem that the existing plastic product micro defect detection device based on super-resolution imaging has a low degree of adaptability between the device detection parameters and the surface defect characteristics of the plastic product, resulting in poor super-resolution imaging quality of the surface defect characteristics, and poor precision and accuracy in the recognition of surface micro defect characteristics. The accuracy of the device detection parameter setting can be improved, thereby improving the super-resolution imaging quality of the surface defect characteristics of the plastic product, achieving the technical goal of accurately identifying the surface defect characteristics, and achieving the technical effect of improving the detection accuracy and quality of micro defects in plastic products.

[0017] Below, the technical solutions in the present application will be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all of the embodiments of the present application. It should be understood that the present application is not limited to the example embodiments described herein. Based on the embodiments of the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present application. It should also be noted that, for the convenience of description, only the parts related to the present application are shown in the accompanying drawings, rather than all of them.

[0018] Embodiment 1

[0019] The present application provides a plastic product micro defect detection device based on super-resolution imaging, the device includes a structured light illumination unit, a microscopic imaging unit, and a self-attention control unit. Figure 1 , the device comprises:

[0020] A feature acquisition module 11 is used to acquire features of the surface to be detected of the plastic product and generate initial detection configuration parameters based on the features of the surface to be detected;

[0021] A response module 12 is used to control the structured light illumination unit to illuminate the surface to be inspected with the initial inspection configuration parameters after the plastic product is positioned;

[0022] An imaging module 13 is used to receive light source through a microscopic imaging unit and establish an initial image;

[0023] A control module 14 is used to call the self-attention control unit, perform imaging detection of initial imaging, and generate a self-attention control result, wherein the self-attention control unit is initialized based on the features of the surface to be detected before detection;

[0024] An adaptive reconstruction module 15 is used to reconstruct the detection configuration parameters based on the self-attention control result, and control the structured light illumination unit and the microscopic imaging unit to re-perform super-resolution imaging according to the reconstructed detection configuration parameters to generate an imaging result;

[0025] The defect detection module 16 is used to receive the imaging result, perform defect detection, and report defect anomalies according to the defect detection result.

[0026] Furthermore, the control module 14 in the device is also used for:

[0027] Inputting the to-be-detected surface features into the self-attention control unit, and performing feature recognition and distribution depth recognition by the self-attention control unit;

[0028] Performing feature similarity clustering based on position continuity according to the feature recognition results to generate a first clustering constraint;

[0029] Based on the distribution depth recognition result, depth similarity clustering based on position continuity is performed to generate a second clustering constraint;

[0030] configuring N adaptive partitions based on the first clustering constraint and the second clustering constraint, where N is an integer greater than 1;

[0031] And based on the features of the surface to be detected, the recognition parameters of N adaptive partitions are configured, the imaging detection of the initial imaging is performed, and the self-attention regulation result is generated.

[0032] Furthermore, the control module 14 in the device is also used for:

[0033] Performing image segmentation of initial imaging through the N adaptive partitions to establish N image blocks;

[0034] Establishing a recognition grayscale median based on the recognition parameters of the N adaptive partitions, and performing grayscale processing on the N image blocks;

[0035] Using the grayscale median to perform binary recognition of the image block after corresponding grayscale processing, and establishing a recognition result;

[0036] The image patch-based focusing result is generated according to the recognition result and is used as the self-attention regulation result.

[0037] Furthermore, the adaptive reconstruction module 15 in the device is also used for:

[0038] generating an attention value of the focusing result based on the recognition result;

[0039] Focus analysis of the attention area is performed based on the attention value and the results of self-attention regulation to generate a tolerant focus space;

[0040] Perform clustering adaptation of focusing results through the tolerant focusing space;

[0041] The focusing result is updated with the clustering adaptation result to complete the reconstruction of the detection configuration parameters.

[0042] Furthermore, the defect detection module 16 in the device is also used for:

[0043] Configure the detection convolution kernel based on the features of the surface to be detected;

[0044] After receiving the imaging result, performing traversal detection of the imaging result through the detection convolution kernel;

[0045] Complete defect detection based on the traversal detection results and generate defect detection results.

[0046] Furthermore, the device includes an optimization module, and the optimization module is used to:

[0047] Perform random inspections of plastic products and create random inspection results using super-resolution imaging;

[0048] Using the sampling results to authenticate the regional detection effect of the non-attention area;

[0049] Optimization parameters are configured according to the authentication result, and imaging module optimization is performed based on the optimization parameters.

[0050] Furthermore, the optimization module in the device is also used for:

[0051] Perform imaging pre-fitting of optimized parameters and generate pre-fitting results;

[0052] Performing fuzzy determination of imaging based on the pre-fitting result to generate a fuzzy determination result;

[0053] M regional imaging focuses are reconstructed based on the fuzzy determination result, and imaging module optimization is performed based on the M regional imaging focuses, where M is an integer greater than 1.

[0054] The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The method for detecting microscopic defects of plastic products based on super-resolution imaging and the specific examples in the second embodiment described later are also applicable to the device for detecting microscopic defects of plastic products based on super-resolution imaging in this embodiment. Through the detailed description of the method for detecting microscopic defects of plastic products based on super-resolution imaging described later, those skilled in the art can clearly understand the device for detecting microscopic defects of plastic products based on super-resolution imaging in this embodiment, so for the sake of brevity of the specification, it will not be described in detail here. For the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the description of the method part.

[0055] Embodiment 2

[0056] Please refer to the attached Figure 2 The present application provides a method for detecting microscopic defects of plastic products based on super-resolution imaging, wherein the method is applied to a device for detecting microscopic defects of plastic products based on super-resolution imaging, the device comprising a structured light illumination unit, a microscopic imaging unit, and a self-attention control unit, and the method specifically comprises the following steps:

[0057] Step 1: Acquire the features of the surface to be inspected of the plastic product, and generate initial inspection configuration parameters based on the features of the surface to be inspected;

[0058] Specifically, the present application provides a method for detecting microscopic defects of plastic products based on super-resolution imaging, which is used to perform quality inspection on microscopic defects on the surface of plastic products to achieve the technical goal of accurately identifying surface defect characteristics and the technical effect of improving the detection accuracy and quality of microscopic defects of plastic products. The method is specifically implemented in a device for detecting microscopic defects of plastic products based on super-resolution imaging. The device includes a structured light illumination unit, a microscopic imaging unit, and a self-attention regulation unit. The structured light illumination unit includes one or more lighting devices, which are used to adjust the light on the surface of the object when imaging a microscopic object, so as to achieve adaptive compensation for the illumination on the surface of the object, thereby improving the imaging quality of the object. The microscopic imaging unit includes a microscopic imaging device, which is used to obtain and process microscopic images of samples. It uses optical principles, electron beams or other imaging technologies to magnify tiny samples to the visible range for detailed observation and analysis; the self-attention regulation unit is used to detect the initial imaging results and obtain imaging quality detection results.

[0059] First, the image of the surface to be inspected of the plastic product is captured by an image sensor, wherein the type of image sensor can be set based on the acquisition accuracy requirements, and the image acquisition result of the surface to be inspected is obtained. Then, feature recognition is performed based on the image acquisition result of the surface to be inspected to determine the features of the surface to be inspected, wherein the features of the surface to be inspected include information such as brightness, color, structure, defect texture, and defect shape. Further, based on the features of the surface to be inspected, the adaptive detection parameter configuration of the micro defect detection equipment is performed to obtain the initial detection configuration parameters, wherein the initial detection configuration parameters include imaging parameters, light source parameters, scanning parameters, etc. The imaging parameters include resolution, focal length and working distance, exposure time, and aperture size, etc. The light source parameters include light source type, light intensity, light angle, light uniformity, etc. The scanning parameters include scanning mode, scanning speed, scanning accuracy, etc. By obtaining the initial detection configuration parameters, support is provided for the subsequent initial imaging of the surface to be inspected.

[0060] Step 2: After the plastic product is positioned, the structured light illumination unit is controlled to illuminate the surface to be inspected with the initial inspection configuration parameters;

[0061] Specifically, when the plastic product is positioned, the surface to be inspected of the plastic product is placed in the area to be imaged; the structured light illumination unit is controlled according to the light source parameters in the initial detection configuration parameters to illuminate the surface to be inspected, thereby providing good lighting conditions for subsequent microscopic imaging of the surface to be inspected.

[0062] Step 3: Receive light source through the microscopic imaging unit and establish initial imaging;

[0063] Specifically, the microscopic imaging unit includes a microscopic imaging device, such as a microscope, etc., which is used to generate and capture a microscopic image of the surface to be inspected. First, the light source is received by the microscopic imaging unit, wherein the light source is an important part of the microscopic imaging unit, which provides the necessary illumination, and the quality and stability of the light source directly affect the clarity and contrast of the image. Then, the surface to be inspected of the plastic product is microscopically imaged by the microscopic imaging device to obtain an initial image of the surface to be inspected.

[0064] Step 4: calling the self-attention control unit to perform imaging detection of initial imaging and generate a self-attention control result, wherein the self-attention control unit is initialized based on the features of the surface to be detected before detection;

[0065] Specifically, the self-attention control unit is then called to perform imaging quality detection on the initial imaging to obtain a self-attention control result, wherein the self-attention control result is a focus detection result obtained after calling the self-attention control unit to detect the initial imaging, wherein the self-attention control unit completes initialization based on the features of the surface to be detected before performing the detection.

[0066] Step 5: Reconstruct the detection configuration parameters based on the self-attention control results, and control the structured light illumination unit and the microscopic imaging unit to re-perform super-resolution imaging according to the reconstructed detection configuration parameters to generate imaging results;

[0067] Specifically, the detection configuration parameters are optimized and adjusted according to the self-attention control results to obtain the reconstructed detection configuration parameters; further, the structured light illumination unit and the microscopic imaging unit are controlled according to the reconstructed detection configuration parameters to re-perform super-resolution imaging on the surface to be detected to obtain imaging results. By optimizing and reconstructing the detection configuration parameters based on the self-attention control results, and performing super-resolution imaging according to the reconstructed detection configuration parameters, the adaptability of the equipment detection parameters to the surface defect characteristics of the plastic product can be improved, thereby improving the super-resolution imaging quality of the surface defect characteristics.

[0068] Step 6: Receive imaging results, perform defect detection, and report defect anomalies based on the defect detection results.

[0069] Specifically, the imaging result is received, and an adapted detection convolution kernel is configured based on the features of the surface to be detected, and then based on a convolutional neural network, the detection convolution kernel is used to identify microscopic defect features on the imaging result to obtain a microscopic defect detection result. The microscopic defect detection result is then judged according to a preset defect constraint, and the preset defect constraint can be set by a person skilled in the art according to the production quality requirements of plastic products, and the defect detection result that does not meet the preset defect constraint is set as a defect abnormality result.

[0070] The method for detecting microscopic defects of plastic products based on super-resolution imaging is applied to a device for detecting microscopic defects of plastic products based on super-resolution imaging, which can solve the technical problems that the existing device for detecting microscopic defects of plastic products based on super-resolution imaging has a low degree of adaptability between the device detection parameters and the surface defect characteristics of the plastic products, resulting in poor super-resolution imaging quality of the surface defect characteristics, and poor precision and accuracy in identifying the surface microscopic defect characteristics. First, the features of the surface to be detected of the plastic product are acquired, and initial detection configuration parameters are generated based on the features of the surface to be detected; then, after the positioning of the plastic product is completed, the structured light illumination unit is controlled by the initial detection configuration parameters to irradiate the light source of the surface to be detected; then, the light source is received by the microscopic imaging unit to establish initial imaging; next, the self-attention regulation unit is called to perform imaging detection of the initial imaging to generate a self-attention regulation result, wherein the self-attention regulation unit is initialized based on the features of the surface to be detected before detection; in addition, the detection configuration parameters are reconstructed based on the self-attention regulation result, and the structured light illumination unit and the microscopic imaging unit are controlled to re-perform super-resolution imaging according to the reconstructed detection configuration parameters to generate imaging results; finally, the imaging results are received, defect detection is performed, and defect abnormalities are reported according to the defect detection results. By collecting the features of the surface to be inspected of the plastic product, the initial inspection parameters are configured based on the features of the surface to be inspected; then the structured light illumination unit and the microscopic imaging unit are called to perform imaging according to the initial inspection parameters to obtain the initial imaging results; then the initial imaging results are inspected for imaging quality by the self-attention control unit to obtain the self-attention control results; further, the inspection parameters are optimized and adjusted based on the self-attention control results, and the inspection configuration parameters are reconstructed to improve the adaptability of the equipment inspection parameters to the surface defect features of the plastic product, and the inspection surface is re-imaged with super-resolution based on the reconstructed inspection configuration parameters to generate imaging results; finally, defect detection is performed on the imaging results to determine the defect abnormality results. The accuracy of the equipment inspection parameter setting can be improved, thereby improving the super-resolution imaging quality of the surface defect features of the plastic product, achieving the technical goal of accurately identifying the surface defect features, and achieving the technical effect of improving the detection accuracy and quality of microscopic defects of plastic products.

[0071] Further, step 4 of this application includes:

[0072] Inputting the to-be-detected surface features into the self-attention control unit, and performing feature recognition and distribution depth recognition by the self-attention control unit;

[0073] Performing feature similarity clustering based on position continuity according to the feature recognition results to generate a first clustering constraint;

[0074] Based on the distribution depth recognition result, depth similarity clustering based on position continuity is performed to generate a second clustering constraint;

[0075] configuring N adaptive partitions based on the first clustering constraint and the second clustering constraint, where N is an integer greater than 1;

[0076] And based on the features of the surface to be detected, the recognition parameters of N adaptive partitions are configured, the imaging detection of the initial imaging is performed, and the self-attention regulation result is generated.

[0077] Specifically, first, the features of the surface to be detected are input into the self-attention control unit, and based on the self-attention control unit, feature recognition and distribution depth recognition are performed on the defect features in the surface to be detected, wherein feature recognition includes defect shape, defect size, etc.; distribution depth recognition refers to the defect depth of the surface to be detected. Then, based on the feature recognition result, defect feature similarity analysis is performed on any two adjacent feature recognition results to obtain feature similarity analysis results, wherein the defect feature similarity analysis can be achieved through the existing cosine similarity comparison method. A feature similarity threshold is set, and any two adjacent feature recognition results that meet the feature similarity threshold are clustered into one category to obtain the first clustering constraint.

[0078] Based on the distributed depth recognition result and the preset depth error threshold, the preset depth error threshold can be set according to the actual situation; and the depth recognition results whose deviation values ​​of any two adjacent depth recognition results are within the preset depth error threshold are clustered into one category to generate a second clustering constraint. Then, N adaptive partitions are configured based on the first clustering constraint and the second clustering constraint to obtain N adaptive partitions, where N is an integer greater than 1. Further, the N adaptive partitions are configured with recognition parameters based on the surface features to be detected. The recognition parameters refer to image quality requirements, including information such as image accuracy and resolution. For example, if the adaptive partition is an area with defective features, the image quality requirements are higher. If the adaptive analysis is an area without defective features, the image quality requirements are lower. The imaging quality detection of the initial imaging is performed according to the recognition parameter configuration result to obtain the self-attention regulation result.

[0079] Furthermore, the present application also includes the following steps:

[0080] Performing image segmentation of initial imaging through the N adaptive partitions to establish N image blocks;

[0081] Establishing a recognition grayscale median based on the recognition parameters of the N adaptive partitions, and performing grayscale processing on the N image blocks;

[0082] Using the grayscale median to perform binary recognition of the image block after corresponding grayscale processing, and establishing a recognition result;

[0083] The image patch-based focusing result is generated according to the recognition result and is used as the self-attention regulation result.

[0084] Specifically, first, the initial imaging is segmented based on the N adaptive partitions to obtain N image blocks. By performing image segmentation, the precision and accuracy of image quality detection can be improved. Then, the grayscale median is established based on the recognition parameters of the N adaptive partitions, that is, first, the grayscale values ​​of all pixels are calculated in each adaptive partition, where the grayscale value reflects the color depth of the point in the image, which can be obtained by converting the RGB value of the color image into a grayscale value. Then, for each adaptive partition, the calculated grayscale values ​​are sorted, and then the median is found, that is, the grayscale value in the middle position. The grayscale median can reflect the overall brightness level of the pixels in the partition, and the grayscale median of the N adaptive partitions is obtained; the N image blocks are further grayscale processed.

[0085] Then, binary recognition of the image block after corresponding grayscale processing is performed according to the grayscale median, wherein the image block binary recognition refers to simplifying the pixel points in the image into only two possible values ​​or grayscale levels. In a binary image, the grayscale value of any pixel point is 0 or 255, representing black and white respectively. This processing method can greatly simplify the image data and improve the processing speed. For example, the grayscale value of the pixel points in the image block greater than the grayscale median is marked as 255, and the grayscale value of the pixel points less than the grayscale median is marked as 0 to obtain the recognition result. Then, the focusing result based on the image block is generated according to the recognition result, that is, the image focusing quality is evaluated according to the recognition result to obtain the image focusing quality evaluation result, and the focusing result is used as the self-attention regulation result.

[0086] Further, step five of this application includes:

[0087] generating an attention value of the focusing result based on the recognition result;

[0088] Focus analysis of the attention area is performed based on the attention value and the results of self-attention regulation to generate a tolerant focus space;

[0089] Perform clustering adaptation of focusing results through the tolerant focusing space;

[0090] The focusing result is updated with the clustering adaptation result to complete the reconstruction of the detection configuration parameters.

[0091] Specifically, an attention value of the focusing result is generated according to the recognition result. The attention value refers to a focusing result evaluation index, which can be set according to actual conditions, such as converting a focusing score or grade into a focusing result evaluation coefficient. These focusing result evaluation coefficients will directly reflect the focusing degree of each image block. Then, a focusing analysis of the attention area is performed according to the attention value and the self-attention regulation result to obtain a tolerant focusing space, which is used to characterize the allowable error range of the focusing result. Then, the focusing results are clustered according to the tolerant focusing space, that is, the focusing results whose adjacent focusing result evaluation coefficients satisfy the tolerant focusing space are clustered into one category. Finally, the focusing result is updated according to the cluster adaptation result, and the reconstruction of the detection configuration parameters is completed based on the updated focusing result.

[0092] Furthermore, step six of this application includes:

[0093] Configure the detection convolution kernel based on the features of the surface to be detected;

[0094] After receiving the imaging result, performing traversal detection of the imaging result through the detection convolution kernel;

[0095] Complete defect detection based on the traversal detection results and generate defect detection results.

[0096] Specifically, the detection convolution kernel is configured based on the features of the surface to be detected. When configuring the detection convolution kernel to adapt to the features of the surface to be detected, we mainly need to focus on the design, training and optimization of the convolution kernel to ensure that it can effectively extract and identify the key features of the surface to be detected. For example: deeply analyze the features of the surface to be detected, such as edges, textures, shapes, etc., and then customize the size, shape and parameters of the detection convolution kernel according to the analysis features. After receiving the imaging results, based on the convolutional neural network, which can be used for image feature extraction, it usually includes multiple convolutional layers, pooling layers, activation functions, and fully connected layers, etc., the imaging results are traversed and detected according to the detection convolution kernel, and defect detection is completed according to the traversal detection results to obtain the defect features of the surface to be detected.

[0097] Furthermore, the present application also includes the following steps:

[0098] Perform random inspections of plastic products and establish random inspection results using super-resolution imaging;

[0099] Using the sampling results to authenticate the regional detection effect of the non-attention area;

[0100] Optimization parameters are configured according to the authentication result, and imaging module optimization is performed based on the optimization parameters.

[0101] Specifically, the device also includes an optimization module. First, a sampling test is performed on the plastic products for surface microscopic defects, and a sampling test result of super-resolution imaging is established based on the sampling test results, wherein the sampling test result includes all super-resolution imaging images. Then, the regional detection effect evaluation is performed on the non-attention area based on the sampling test results, wherein the non-attention area refers to the area that does not contain surface defect features, and the detection effect evaluation result of the non-attention area is obtained, that is, the certification result. Further, the optimization parameters are configured according to the certification result. For example, when the detection effect evaluation result does not meet the expected indicators, the device detection parameters can be optimized to improve the detection quality of the non-attention area. Finally, the imaging module is optimized according to the optimization parameters.

[0102] Furthermore, the present application also includes the following steps:

[0103] Perform imaging pre-fitting of optimized parameters and generate pre-fitting results;

[0104] Performing fuzzy determination of imaging based on the pre-fitting result to generate a fuzzy determination result;

[0105] M regional imaging focuses are reconstructed based on the fuzzy determination result, and imaging module optimization is performed based on the M regional imaging focuses, where M is an integer greater than 1.

[0106] Specifically, first, microscopic imaging is performed on the surface defects of the plastic product based on the optimization parameters to obtain an imaging pre-fitting result. Then, blur judgment of the imaging is performed based on the pre-fitting result, wherein blur judgment refers to evaluating the image blur or clarity of the pre-fitting result to obtain a blur judgment result. Finally, M regional imaging focuses are reconstructed according to the blur judgment result, wherein M is an integer greater than 1. First, the blur judgment result is analyzed to determine which areas in the image are judged to be blurred; then, for each area to be reconstructed, the imaging focus is recalculated or adjusted according to the physical characteristics and optical principles of the imaging system; finally, the imaging module is optimized according to the M regional imaging focuses.

[0107] By constructing an optimization module to analyze the imaging quality of the non-attention area, reconstructing the regional imaging focus according to the imaging quality analysis results of the non-attention area, and optimizing the imaging module, the microscopic imaging quality of the surface to be detected can be further improved.

[0108] In summary, the method for detecting microscopic defects of plastic products based on super-resolution imaging provided by the present application has the following technical effects:

[0109] 1. By collecting the features of the surface to be inspected of the plastic product, the initial inspection parameters are configured based on the features of the surface to be inspected; then the structured light illumination unit and the microscopic imaging unit are called to perform imaging according to the initial inspection parameters to obtain the initial imaging results; then the self-attention control unit performs imaging quality inspection on the initial imaging results to obtain the self-attention control results; further based on the self-attention control results, the inspection parameters are optimized and adjusted, and the inspection configuration parameters are reconstructed to improve the adaptability of the equipment inspection parameters to the surface defect features of the plastic product, and the inspection surface is re-imaged with super-resolution based on the reconstructed inspection configuration parameters to generate imaging results; finally, defect detection is performed on the imaging results to determine the defect abnormality results. The accuracy of the equipment inspection parameter setting can be improved, thereby improving the super-resolution imaging quality of the surface defect features of the plastic product, achieving the technical goal of accurately identifying the surface defect features, and achieving the technical effect of improving the detection accuracy and quality of microscopic defects of plastic products.

[0110] 2. By optimizing and reconstructing the detection configuration parameters based on the self-attention control results, and performing super-resolution imaging according to the reconstructed detection configuration parameters, the adaptability of the equipment detection parameters to the surface defect characteristics of plastic products can be improved, thereby improving the super-resolution imaging quality of surface defect characteristics.

[0111] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

[0112] Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application belong to the scope of the present application and its equivalent technology, the present application is also intended to include these modifications and variations.

Claims

1. A plastic product micro defect detection device based on super-resolution imaging, the device comprising a structured light illumination unit, a microscopic imaging unit, and a self-attention control unit, It is characterized in that The device comprises: A feature acquisition module, used to acquire features of the surface to be inspected of the plastic product, and generate initial inspection configuration parameters based on the features of the surface to be inspected; A response module is used to control the structured light illumination unit to illuminate the surface to be inspected with the initial inspection configuration parameters after the plastic product is positioned; An imaging module, used for receiving light source through a microscopic imaging unit and establishing initial imaging; A control module, used for calling the self-attention control unit, performing imaging detection of initial imaging, and generating a self-attention control result, wherein the self-attention control unit is initialized based on the features of the surface to be detected before detection; An adaptive reconstruction module is used to reconstruct the detection configuration parameters based on the self-attention control results, and control the structured light illumination unit and the microscopic imaging unit to re-perform super-resolution imaging according to the reconstructed detection configuration parameters to generate imaging results; A defect detection module is used to receive imaging results, perform defect detection, and report defect anomalies based on the defect detection results; Wherein, the control module is also used for: Inputting the to-be-detected surface features into the self-attention control unit, and performing feature recognition and distribution depth recognition by the self-attention control unit; Performing feature similarity clustering based on position continuity according to the feature recognition results to generate a first clustering constraint; Based on the distribution depth recognition result, depth similarity clustering based on position continuity is performed to generate a second clustering constraint; configuring N adaptive partitions based on the first clustering constraint and the second clustering constraint, where N is an integer greater than 1; and configuring recognition parameters of N adaptive partitions based on the features of the surface to be detected, performing imaging detection of the initial imaging, and generating the self-attention regulation result; Wherein, the control module is also used for: Performing image segmentation of initial imaging through the N adaptive partitions to establish N image blocks; Establishing a recognition grayscale median based on the recognition parameters of the N adaptive partitions, and performing grayscale processing on the N image blocks; Using the grayscale median to perform binary recognition of the image block after corresponding grayscale processing, and establishing a recognition result; Generate a focusing result based on the image block according to the recognition result, and use it as the self-attention regulation result; Wherein, the adaptive reconstruction module is also used for: generating an attention value of the focusing result based on the recognition result; Focus analysis of the attention area is performed based on the attention value and the results of self-attention regulation to generate a tolerant focus space; Perform clustering adaptation of focusing results through the tolerant focusing space; The focusing result is updated with the clustering adaptation result to complete the reconstruction of the detection configuration parameters.

2. The device as claimed in claim 1, It is characterized in that The defect detection module is also used for: Configure the detection convolution kernel based on the features of the surface to be detected; After receiving the imaging result, performing traversal detection of the imaging result through the detection convolution kernel; Complete defect detection based on the traversal detection results and generate defect detection results.

3. The device as claimed in claim 1, It is characterized in that The device also includes an optimization module, which is used to: Perform random inspections of plastic products and create random inspection results using super-resolution imaging; Using the sampling results to authenticate the regional detection effect of the non-attention area; Optimization parameters are configured according to the authentication result, and imaging module optimization is performed based on the optimization parameters.

4. The device as claimed in claim 3, It is characterized in that The optimization module is also used to: Perform imaging pre-fitting of optimized parameters and generate pre-fitting results; Performing fuzzy determination of imaging based on the pre-fitting result to generate a fuzzy determination result; M regional imaging focuses are reconstructed based on the fuzzy determination result, and imaging module optimization is performed based on the M regional imaging focuses, where M is an integer greater than 1.

5. A method for detecting microscopic defects of plastic products based on super-resolution imaging. It is characterized in that Steps for implementing the device of any one of claims 1 to 4, the device comprising a structured light illumination unit, a microscopic imaging unit, and a self-attention regulation unit, the method comprising: Acquire the features of the surface to be inspected of the plastic product, and generate initial inspection configuration parameters based on the features of the surface to be inspected; When the plastic product is positioned, the structured light illumination unit is controlled to illuminate the surface to be inspected with the initial inspection configuration parameters; Receive light source through the microscopic imaging unit to establish initial imaging; Calling the self-attention regulation unit to perform imaging detection of initial imaging and generate a self-attention regulation result, wherein the self-attention regulation unit is initialized based on the features of the surface to be detected before detection; Reconstruct the detection configuration parameters based on the self-attention control results, and control the structured light illumination unit and the microscopic imaging unit to re-perform super-resolution imaging according to the reconstructed detection configuration parameters to generate imaging results; Receive imaging results, perform defect detection, and report defect anomalies based on the defect detection results.

Images