Feature classification-based welding spot defect infrared detection method and system

Abstract

The invention relates to the technical field of infrared detection, in particular to a welding spot defect infrared detection method and system based on feature classification, and the method comprises the steps: detecting a processed product through an infrared thermal imager to obtain an infrared image, positioning the collected infrared image to determine a welding spot region, and determining the welding spot defect by using a feature classification method; an infrared image containing a welding spot area is screened out for preprocessing, and an input infrared image is generated; comparing the input infrared image with a complete infrared image in the infrared image complete dictionary, detecting whether a welding spot defect exists or not, and generating first detection information if the welding spot defect exists; performing welding spot thermal feature extraction on the input infrared image corresponding to the generated first detection information, inputting the extracted welding spot thermal features into a welding spot detection classifier to judge a defect type, and generating second detection information according to the defect type; searching and sending an information code in a database according to the second detection information; the accuracy and efficiency of infrared detection can be improved at the same time.

Description

Technical Field

[0001] This invention relates to the field of infrared detection technology, and in particular to an infrared detection method and system for weld joint defects based on feature classification. Background Technology

[0002] In modern industrial production, the use of infrared thermal imaging for weld defect detection has become an urgent need. Currently, in the field of infrared detection technology, most research focuses on solving the problems of accuracy or efficiency in infrared detection, with few methods addressing the simultaneous improvement of both accuracy and efficiency.

[0003] For example, Chinese patent CN117740831B discloses a semiconductor chip welding quality analysis system based on infrared vision. This technical solution collects data on the circuit board itself and the solder joints through infrared vision recognition, performs defect analysis on the surface of the solder joints, and analyzes the three-dimensional solid parameters using coordinate analysis, 3D modeling, and data comparison analysis. This allows for a specific determination of whether the solder joint specifications are appropriate, whether there is too much or too little solder, whether the pins are completely covered, and whether the circuitry on the back of the circuit board is touched. This comprehensive analysis ensures that the solder joints meet the standards, effectively guaranteeing the welding quality of the semiconductor chip and thus improving the overall quality of the circuit board.

[0004] For example, Chinese patent CN115452888B discloses a solder joint quality inspection equipment and method based on infrared thermography. This technical solution collects the temperature-time curve of the solder joint under pulsed thermal excitation, uses a solder joint defect feature parameter extraction algorithm to output feature parameters characterizing the solder joint quality, and further uses a solder joint defect judgment model to output the defect area percentage of the solder joint.

[0005] All of the above technical solutions suffer from the problem that they cannot simultaneously improve the accuracy and efficiency of infrared detection. Summary of the Invention

[0006] The main objective of this invention is to provide a method and system for infrared detection of weld defects based on feature classification, which effectively solves the problems mentioned in the background art.

[0007] The technical solution of the present invention is as follows:

[0008] Firstly, a feature-based infrared detection method for solder joint defects is proposed, which includes the following steps:

[0009] S1. The product after processing and pre-excitation by an external heat source is detected by an infrared thermal imager. The infrared image is obtained based on the difference in surface temperature field distribution between the defect area and the normal area after thermal excitation. The acquired infrared image is used to locate and determine the solder joint area. The infrared image containing the solder joint area is selected for preprocessing and the input infrared image is generated.

[0010] S2. Compare the input infrared image with the complete infrared images in the infrared image dictionary to detect whether there are solder joint defects. If there are solder joint defects, generate the first detection information.

[0011] S3. Extract the thermal features of the solder joints from the input infrared image corresponding to the first detection information, input the extracted thermal features of the solder joints into the solder joint detection classifier to determine the defect type, and generate the second detection information based on the defect type.

[0012] S4. Based on the second detection information, search the database and send the information code.

[0013] A further improvement of the present invention is that the preprocessing in S1 includes the following specific steps:

[0014] S101: Convert the selected infrared image containing the solder joint area into a grayscale image;

[0015] S102: Extract perceptual information from the grayscale image to generate the first source image;

[0016] S103: Use a filter to filter and decompose the grayscale image to remove noise interference information, and use the decomposed texture image as the second source image;

[0017] S104: Perform image fusion between the first source image and the second source image to generate the input infrared image.

[0018] A further improvement of the present invention is that the infrared image complete dictionary in S2 is an infrared image database formed by randomized samples after infrared image acquisition and processing, and the samples in the image database are all complete infrared images without solder joint defects.

[0019] A further improvement of the present invention is that the extraction of the thermal features of the solder joint in S3 includes the following specific steps:

[0020] S301: Based on the input infrared image, count the number of connected regions of thermal anomalies between two adjacent solder joints;

[0021] S302: Statistically analyze the characteristic information of the thermal anomaly connected region of the solder joint, including the centroid location and perimeter of the thermal anomaly connected region. Area of ​​the thermally abnormal connection zone of the solder joint .

[0022] A further improvement of the present invention is that the statistical analysis of the characteristic information of the thermal anomaly connectivity region of the solder joint in S302 includes the following specific steps:

[0023] Optionally, the extraction of the thermal features of the solder joints in S3 includes the following specific steps:

[0024] S301: Based on the input infrared image, count the number of connected regions of thermal anomalies between two adjacent solder joints;

[0025] S302: Statistically analyze the characteristic information of the thermal anomaly connected region of the solder joint, including the centroid location and perimeter of the thermal anomaly connected region. Area of ​​the thermally abnormal connection zone of the solder joint .

[0026] Optionally, the statistical analysis of the characteristic information of the thermal anomaly connectivity region of the solder joint in S302 includes the following specific steps:

[0027] S3021: Extract the centroid position of the thermal anomaly connected region of the solder joint as the center coordinate, and establish a planar coordinate system with the center coordinate as the origin to define the pixel coordinates within a single thermal anomaly region;

[0028] S3022: Calculate the perimeter of the thermal anomaly connected region of the solder joint, wherein the thermal anomaly connected region of the solder joint is the pixel area with a preset gray value in the thermal anomaly region of a single solder joint in the input infrared image.

[0029] Optionally, calculate the perimeter of the thermal anomaly connected region of the solder joint, specifically including:

[0030] Determine the x-coordinate and y-coordinate ranges of the boundary pixels of the thermal anomaly connected region of the solder joint;

[0031] Pixels that are within the range of the horizontal and vertical coordinates and belong to the boundary of the region are selected by using the pixel function of the infrared image of the thermal anomaly area of ​​a single solder joint.

[0032] The total number of selected boundary pixels is counted, and the perimeter and area of ​​the thermal anomaly connected region of the solder joint are determined based on this total number.

[0033] A further improvement of the present invention is that the implementation of the solder joint detection classifier in S3 includes the following specific steps:

[0034] S311: Acquire infrared sample images of weld joint defects and classify the defects from the infrared sample images;

[0035] S312: The infrared sample images are matched one-to-one with the classification results to form training units, and the set of training units is used as the training set.

[0036] S313: Using the thermal features of the solder joints in the infrared sample images as input data, a convolutional neural network is used to train the features of the training set, outputting a template for judging the features of solder joint defects and forming a solder joint detection classifier.

[0037] A further improvement of the present invention is that S313 includes:

[0038] S3131: If the number of connected areas of thermal abnormality is 2 or 3, it is determined that there is a solder bridging defect. When there is only one connected area of ​​thermal abnormality, it is a solder joint adhesion.

[0039] S3132: Given a bridging detection threshold When the perimeter of the connected area of ​​the solder joint is abnormally hot If so, a bridging defect is determined to exist;

[0040] S3133: Given the offset detection threshold When the area of ​​the connected region of the solder joint is abnormally hot within the confidence interval Inside, and center coordinates If the coordinates of the template center generated by the solder joint detection classifier do not match, it is determined that the solder joint is misaligned.

[0041] A further improvement of the present invention is that the database in S4 contains thermal characteristic data of different solder joint defect types, and generates a unique information code according to different solder joint defect types.

[0042] Secondly, a feature-based infrared detection system for weld joint defects is proposed, which includes:

[0043] Data acquisition and processing module, comparison and filtering module, feature classification module, and search and identification module;

[0044] The acquisition and processing module is used to acquire infrared images of the processed product, locate and determine the solder joint area from the acquired infrared images, filter out the infrared images containing the solder joint area for preprocessing, and generate the input infrared image.

[0045] The comparison and screening module is used to compare the input infrared image with the complete infrared images in the infrared image complete dictionary to detect whether there are solder joint defects. If there are solder joint defects, first detection information is generated.

[0046] The feature classification module is used to extract the thermal features of the solder joints from the input infrared image corresponding to the first detection information, input the extracted thermal features of the solder joints into the solder joint detection classifier to determine the defect type, and generate the second detection information based on the defect type.

[0047] The search identifier module is used to search the database and send an information code based on the second detection information.

[0048] The technical effects of this invention are as follows:

[0049] This invention constructs an infrared detection method for solder joint defects based on feature classification, overcoming the problems of low efficiency, long processing time, and low accuracy in the solder joint defect classification process. It effectively improves the efficiency of solder joint defect classification, achieving accurate and efficient classification. This method involves capturing and processing infrared images of thermally excited products, extracting thermal features from the infrared images containing solder joint defects, and then using the samples for feature training to form a solder joint detection classifier. The classifier is then directly used to classify defect types. Therefore, it offers good operability, high accuracy, good stability, and high flexibility, and has significant practical implications for improving the efficiency and accuracy of solder joint defect classification. Attached Figure Description

[0050] Other features, objects, and advantages of the invention will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings:

[0051] Figure 1 This is a flowchart illustrating an infrared detection method for solder joint defects based on feature classification, according to Embodiment 1 of the present invention.

[0052] Figure 2 This is a schematic diagram of the image processing flow of an infrared detection method for solder joint defects based on feature classification according to Embodiment 1 of the present invention;

[0053] Figure 3 This is a schematic diagram of the structure of an infrared detection system for weld defects based on feature classification, according to Embodiment 2 of the present invention. Detailed Implementation

[0054] Example 1

[0055] This embodiment discloses an infrared detection method for solder joint defects based on feature classification, such as... Figure 1 , Figure 2 As shown, the specific steps include the following:

[0056] S1. The product after processing and pre-excitation by an external heat source is detected by an infrared thermal imager. The infrared image is obtained based on the difference in surface temperature field distribution between the defect area and the normal area after thermal excitation. The acquired infrared image is used to locate and determine the solder joint area. The infrared image containing the solder joint area is selected for preprocessing and the input infrared image is generated.

[0057] In this embodiment, the specific steps for preprocessing the infrared image containing the solder joint area are as follows:

[0058] S101: Convert the selected infrared image containing the solder joint area into a grayscale image;

[0059] S102: The first source image is generated by extracting the perceptual detail information of the grayscale image using the VGG16 network. The infrared image contains rich perceptual detail information and can effectively capture the subtle temperature distribution patterns caused by defects in the infrared image, which can enhance the perceptual details in subsequent infrared image fusion.

[0060] S103: Due to the influence of infrared thermal imager noise, uneven thermal excitation, environmental radiation interference and other factors, the actual infrared images will be mixed with various noise interference information. Before analyzing the infrared images, it is necessary to use a filter to filter and decompose the grayscale images, remove noise interference information, retain the temperature boundary caused by the real defects, and obtain a texture image containing thermal structure edge contour information after decomposition. The texture image is used as the second source image.

[0061] S104: Perform image fusion between the first source image and the second source image to improve the detail quality of the second source image and generate the input image.

[0062] S2. Compare the input infrared image with the complete infrared images in the infrared image dictionary to detect whether there are solder joint defects. If there are solder joint defects, generate the first detection information.

[0063] In this embodiment, the infrared image complete dictionary is an infrared image database formed by randomized samples after infrared image acquisition and processing. The samples in the infrared image database are all complete infrared images without solder joint defects. By comparing the complete infrared images with the input infrared images, the input infrared images without solder joint defects can be filtered out, and the infrared images with solder joint defects can be selected.

[0064] S3. Extract the thermal features of the solder joints from the input infrared image corresponding to the first detection information, input the extracted thermal features of the solder joints into the solder joint detection classifier to determine the defect type, and generate the second detection information based on the defect type.

[0065] In this embodiment, the specific steps for extracting the thermal features of the solder joints are as follows:

[0066] S301: Based on the input infrared image, count the number of connected regions of thermal anomalies between two adjacent solder joints;

[0067] S302: Statistically analyze the characteristic information of the thermal anomaly connected region of the solder joint, including the centroid location and perimeter of the thermal anomaly connected region. Area of ​​the thermally abnormal connection zone of the solder joint .

[0068] In this embodiment, the specific steps for statistically analyzing the characteristic information of the thermal anomaly connectivity region of the solder joint are as follows:

[0069] S3021: Extract the centroid position of the thermal anomaly connected region of the solder joint as the center coordinate. A planar coordinate system is established with the center coordinates as the origin to define the pixel coordinates within a single thermal anomaly region;

[0070] S3022: Calculate the perimeter of the thermal anomaly connected region of the solder joint. The thermal anomaly connected region of the solder joint is the pixel area with a preset grayscale value in the thermal anomaly region of a single solder joint in the input infrared image. The perimeter of the thermal anomaly connected region of the solder joint is the total number of boundary pixels of the pixel area with the preset grayscale value. The calculation formula is as follows:

[0071] ;

[0072] in, , , The perimeter of the thermally abnormal connected area of ​​the solder joint. The x-coordinate of the infrared image pixel in the thermal anomaly region of a single solder joint. The vertical coordinate of the infrared image pixel in the thermal anomaly region of a single solder joint. This is a pixel function of the infrared image for a single solder joint thermal anomaly region. This represents the minimum x-coordinate of the boundary pixel of the thermal anomaly connected region of the solder joint. This represents the minimum ordinate of the pixel at the boundary of the thermal anomaly connected region of the solder joint. This represents the maximum x-coordinate of the pixel at the boundary of the thermal anomaly connected region of the solder joint. This represents the maximum value of the ordinate of the pixel at the boundary of the thermally abnormal connected region of the solder joint;

[0073] S3023: Calculate the area of ​​the thermal anomaly connected region of the solder joint. The calculation formula is as follows:

[0074] ;

[0075] in, , .

[0076] In this embodiment, the implementation of the solder joint detection classifier includes the following specific steps:

[0077] S311: Acquire infrared sample images of weld joint defects and classify the defects from the infrared sample images;

[0078] S312: The infrared sample images are matched one-to-one with the classification results to form training units, and the set of training units is used as the training set.

[0079] S313: Using the thermal features of the solder joints in the infrared sample images as input data, a convolutional neural network is used to train the features of the training set, outputting a template for judging the features of solder joint defects and forming a solder joint detection classifier.

[0080] In this embodiment, S313 includes:

[0081] S3131: If the number of connected areas of thermal abnormality is 2 or 3, it is determined that there is a solder bridging defect. In particular, when there is only one connected area of ​​thermal abnormality, it is solder joint adhesion.

[0082] S3132: Given a bridging detection threshold When the perimeter of the connected area of ​​the solder joint is abnormally hot When a bridging defect is detected, the bridging detection threshold is set. The specific application requirements shall be determined by those skilled in the art.

[0083] S3133: For normal solder joint infrared images, the area of ​​a given bridging detection threshold will remain within a certain confidence interval. However, for misaligned solder joints, the center coordinates of the thermal anomaly connected region will differ significantly from the center coordinates of the template. Therefore, a given misalignment detection threshold... When the area of ​​the connected region of the solder joint is abnormally hot within the confidence interval Inside, and center coordinates When the coordinates of the template center generated by the solder joint detection classifier do not match, it is determined that the solder joint is misaligned. The misalignment detection threshold is... The specific application requirements shall be determined by those skilled in the art.

[0084] S4. Based on the second detection information, search the database and send the information code.

[0085] In this embodiment, the database contains thermal characteristic data of different solder joint defect types, which can be automatically searched and matched according to the second monitoring information to generate a unique information code.

[0086] Example 2

[0087] This embodiment proposes an infrared detection system for solder joint defects based on feature classification, such as... Figure 3 As shown, it includes: a data acquisition and processing module, a comparison and filtering module, a feature classification module, and a search and identification module;

[0088] The acquisition and processing module is used to acquire infrared images of the processed product, locate and determine the solder joint area from the acquired infrared images, filter out the infrared images containing the solder joint area for preprocessing, and generate the input infrared image.

[0089] In this embodiment, the preprocessing of the infrared image containing the solder joint area includes the following specific steps: First, the acquired infrared image containing the solder joint area is converted into a grayscale image, and a VGG16 network is used to extract the perceptual detail information of the grayscale image to generate a first source image. The perceptual infrared image contains rich perceptual detail information, which can effectively capture the subtle temperature distribution patterns caused by defects in the infrared image, and can enhance the perceptual details in subsequent infrared image fusion. Then, a filter is used to filter and decompose the grayscale image to remove noise interference information and retain the temperature boundary caused by the real defect. After decomposition, a texture image containing thermal structure edge contour information is obtained, and the texture image is used as the second source image. Finally, the first source image and the second source image are fused to improve the detail quality of the second source image and generate the input infrared image.

[0090] The comparison and screening module is used to compare the input infrared image with the complete infrared images in the infrared image complete dictionary to detect whether there are solder joint defects. If there are solder joint defects, first detection information is generated.

[0091] In this embodiment, the infrared image complete dictionary is an infrared image database formed by randomized samples after infrared image acquisition and processing. The samples in the infrared image database are all complete infrared images without solder joint defects. By comparing the complete infrared images with the input infrared images, the input infrared images without solder joint defects can be filtered out, and the infrared images with solder joint defects can be selected.

[0092] The feature classification module is used to extract the thermal features of the solder joints from the input infrared image corresponding to the first detection information, input the extracted thermal features of the solder joints into the solder joint detection classifier to determine the defect type, and generate the second detection information based on the defect type.

[0093] In this embodiment, the solder joint feature extraction includes the following specific implementation steps: First, based on the input infrared image, the number of connected regions of thermal anomalies between two adjacent solder joints is determined; then, the feature information of the connected regions of thermal anomalies is counted, including the centroid position and perimeter of the connected regions of thermal anomalies. and area The specific implementation steps for statistically analyzing the characteristics of the thermal anomaly connectivity region at solder joints include the following:

[0094] Optionally, the extraction of the thermal features of the solder joints in S3 includes the following specific steps:

[0095] S301: Based on the input infrared image, count the number of connected regions of thermal anomalies between two adjacent solder joints;

[0096] S302: Statistically analyze the characteristic information of the thermal anomaly connected region of the solder joint, including the centroid location and perimeter of the thermal anomaly connected region. Area of ​​the thermally abnormal connection zone of the solder joint .

[0097] Optionally, the statistical analysis of the characteristic information of the thermal anomaly connectivity region of the solder joint in S302 includes the following specific steps:

[0098] S3021: Extract the centroid position of the thermal anomaly connected region of the solder joint as the center coordinate, and establish a planar coordinate system with the center coordinate as the origin to define the pixel coordinates within a single thermal anomaly region;

[0099] S3022: Calculate the perimeter of the thermal anomaly connected region of the solder joint, wherein the thermal anomaly connected region of the solder joint is the pixel area with a preset gray value in the thermal anomaly region of a single solder joint in the input infrared image.

[0100] Optionally, calculate the perimeter of the thermal anomaly connected region of the solder joint, specifically including:

[0101] Determine the x-coordinate and y-coordinate ranges of the boundary pixels of the thermal anomaly connected region of the solder joint;

[0102] Pixels that fall within the above-mentioned horizontal and vertical coordinate range and belong to the region boundary are selected by using the pixel function of the infrared image of the thermal anomaly area of ​​a single solder joint.

[0103] The total number of selected boundary pixels is counted, and the perimeter and area of ​​the solder joint thermal anomaly connected region are determined based on this total number. Specifically, the centroid position of the solder joint thermal anomaly connected region is first extracted as the center coordinate. A small planar coordinate system is established with the center coordinates as the origin to define the pixel coordinates within a single thermal anomaly connected region. Then, the perimeter of the solder joint thermal anomaly connected region is calculated. This region is defined as the pixel region with a preset grayscale value within a single solder joint thermal anomaly region of the input infrared image. The perimeter of the solder joint thermal anomaly connected region is the total number of boundary pixels of the pixel region with the preset grayscale value. The calculation formula is as follows:

[0104] ;

[0105] in, , , The perimeter of the thermally abnormal connected area of ​​the solder joint. The x-coordinate of the infrared image pixel in the thermal anomaly region of a single solder joint. The vertical coordinate of the infrared image pixel in the thermal anomaly region of a single solder joint. This is a pixel function of the infrared image for a single solder joint thermal anomaly region. The expression is:

[0106] ;

[0107] B is the set of pixels that form the boundary of a pixel region with a preset grayscale value. This represents the minimum x-coordinate of the boundary pixel of the thermal anomaly connected region of the solder joint. This represents the minimum ordinate of the pixel at the boundary of the thermal anomaly connected region of the solder joint. This represents the maximum x-coordinate of the pixel at the boundary of the thermal anomaly connected region of the solder joint. The maximum value of the ordinate of the boundary pixel of the thermal anomaly connected region of the solder joint is given. Finally, the area of ​​the thermal anomaly connected region of the solder joint is calculated by counting the number of pixels in the region, i.e., the total number of pixels within the boundary. The calculation formula is as follows:

[0108] ;

[0109] in, , .

[0110] In this embodiment, the implementation of the solder joint detection classifier includes the following specific steps: First, infrared sample images of solder joint defects are acquired, and the infrared sample images are classified as defects; then, the infrared sample images and classification results are matched one-to-one to form training units, and the set of training units is used as the training set; finally, the thermal features of the solder joints in the infrared sample images are used as input data, and a convolutional neural network is used to train the features of the training set, outputting a template for judging the features of solder joint defects and forming a solder joint detection classifier. The specific content of the solder joint detection classifier is as follows:

[0111] 1) If the number of connected areas of thermal abnormality is 2 or 3, it is determined that there is a solder bridging defect. In particular, when there is only one connected area of ​​thermal abnormality, it is solder joint adhesion.

[0112] 2) Given a bridging detection threshold When the perimeter of the connected area of ​​the solder joint is abnormally hot If a bridging defect is detected, the bridging detection threshold is determined to be present. The specific application requirements shall be determined by those skilled in the art.

[0113] 3) For normal solder joint infrared images, the area of ​​the thermal anomaly connected region will remain within a certain confidence interval. However, for misaligned solder joints, the center coordinates of the thermal anomaly connected region will differ significantly from the center coordinates of the template. Therefore, a misalignment detection threshold is given. When the area of ​​the connected region of the solder joint is abnormally hot within the confidence interval Inside, and the center coordinates When the coordinates do not match the center coordinates of the template, it is determined that the solder joint is misaligned. The misalignment detection threshold is... The specific application requirements shall be determined by those skilled in the art.

[0114] The search identifier module is used to search the database and send an information code based on the second detection information.

[0115] In this embodiment, the database contains thermal characteristic data of different solder joint defect types, which can be automatically searched and matched according to the second monitoring information to generate a unique information code.

[0116] The parameters and steps for implementing the corresponding functions of each unit module in the infrared detection system for weld joint defects based on feature classification of the present invention can be referred to the parameters and steps in the embodiment of the infrared detection method for weld joint defects based on feature classification in Embodiment 1 above.

[0117] Those skilled in the art will recognize that this invention can be implemented as a system, method, or computer program product. Therefore, this disclosure can be embodied in the following forms: it can be entirely hardware, entirely software (including firmware, resident software, microcode, etc.), or a combination of hardware and software, generally referred to herein as a "circuit," "module," or "system." Furthermore, in some embodiments, the invention can also be implemented as a computer program product contained in one or more computer-readable media, which includes computer-readable program code.

[0118] Any combination of one or more computer-readable media may be used. A computer-readable medium can be a computer-readable signal medium or a computer-readable storage medium. A computer-readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples (a non-exhaustive list) of computer-readable storage media include: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this document, a computer-readable storage medium can be any tangible medium that contains or stores a program that can be used by or in connection with an instruction execution system, apparatus, or device.

[0119] This invention is described with reference to flowchart illustrations and block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and block diagrams, as well as combinations of blocks in the flowchart illustrations and block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart. Figure 1 One or more processes and boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0120] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and boxes Figure 1 The steps of the function specified in one or more boxes.

[0121] The embodiments of the present invention have been described above with reference to the accompanying drawings. However, the present invention is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms under the guidance of the present invention without departing from the spirit and scope of the claims. All of these forms are within the protection scope of the present invention.

Claims

1. A method for infrared detection of weld joint defects based on feature classification, characterized in that: The specific steps include the following: S1. The product after processing and pre-excitation by an external heat source is detected by an infrared thermal imager. The infrared image is obtained based on the difference in surface temperature field distribution between the defect area and the normal area after thermal excitation. The acquired infrared image is used to locate and determine the solder joint area. The infrared image containing the solder joint area is selected for preprocessing and the input infrared image is generated. S2. Compare the input infrared image with the complete infrared images in the infrared image dictionary to detect whether there are solder joint defects. If there are solder joint defects, generate the first detection information. S3. Extract the thermal features of the solder joints from the input infrared image corresponding to the first detection information, input the extracted thermal features of the solder joints into the solder joint detection classifier to determine the defect type, and generate the second detection information based on the defect type. S4. Based on the second detection information, search the database and send the information code.

2. The infrared detection method for solder joint defects based on feature classification according to claim 1, characterized in that: The preprocessing in S1 includes the following specific steps: S101: Convert the selected infrared image containing the solder joint area into a grayscale image; S102: Extract perceptual information from the grayscale image to generate the first source image; S103: Use a filter to filter and decompose the grayscale image to remove noise interference information, and use the decomposed texture image as the second source image; S104: Perform image fusion between the first source image and the second source image to generate the input infrared image.

3. The infrared detection method for solder joint defects based on feature classification according to claim 1, characterized in that: The infrared image complete dictionary in S2 is an infrared image database formed by the acquisition and processing of randomized samples. All samples in the image database are complete infrared images without solder joint defects.

4. The infrared detection method for weld joint defects based on feature classification according to claim 1, characterized in that, The extraction of the thermal features of the solder joints in S3 includes the following specific steps: S301: Based on the input infrared image, count the number of connected regions of thermal anomalies between two adjacent solder joints; S302: Statistically analyze the characteristic information of the thermal anomaly connected region of the solder joint, including the centroid location and perimeter of the thermal anomaly connected region. Area of ​​the thermally abnormal connection zone of the solder joint .

5. The infrared detection method for solder joint defects based on feature classification according to claim 4, characterized in that: The statistical analysis of the characteristic information of the thermal anomaly connectivity region of the solder joint in S302 includes the following specific steps: S3021: Extract the centroid position of the thermal anomaly connected region of the solder joint as the center coordinate, and establish a planar coordinate system with the center coordinate as the origin to define the pixel coordinates within a single thermal anomaly region; S3022: Calculate the perimeter of the thermal anomaly connected region of the solder joint, wherein the thermal anomaly connected region of the solder joint is the pixel area with a preset gray value in the thermal anomaly region of a single solder joint in the input infrared image.

6. The infrared detection method for solder joint defects based on feature classification according to claim 5, characterized in that: Calculating the perimeter of the thermal anomaly connected region of the solder joint specifically includes: Determine the x-coordinate and y-coordinate ranges of the boundary pixels of the thermal anomaly connected region of the solder joint; Pixels that are within the range of the horizontal and vertical coordinates and belong to the boundary of the region are selected by using the pixel function of the infrared image of the thermal anomaly area of ​​a single solder joint. The total number of selected boundary pixels is counted, and the perimeter and area of ​​the thermal anomaly connected region of the solder joint are determined based on this total number.

7. The infrared detection method for solder joint defects based on feature classification according to claim 5, characterized in that: The implementation of the solder joint detection classifier in S3 includes the following specific steps: S311: Acquire infrared sample images of weld joint defects and classify the defects from the infrared sample images; S312: The infrared sample images are matched one-to-one with the classification results to form training units, and the set of training units is used as the training set. S313: Using the thermal features of the solder joints in the infrared sample images as input data, a convolutional neural network is used to train the features of the training set, outputting a template for judging the features of solder joint defects and forming a solder joint detection classifier.

8. The infrared detection method for solder joint defects based on feature classification according to claim 7, characterized in that: S313 includes: S3131: If the number of connected areas of thermal abnormality is 2 or 3, it is determined that there is a solder bridging defect. When there is only one connected area of ​​thermal abnormality, it is a solder joint adhesion. S3132: Given a bridging detection threshold When the perimeter of the connected area of ​​the solder joint is abnormally hot If so, a bridging defect is determined to exist; S3133: Given the offset detection threshold When the area of ​​the connected region of the solder joint is abnormally hot within the confidence interval Inside, and center coordinates If the coordinates of the template center generated by the solder joint detection classifier do not match, it is determined that the solder joint is misaligned.

9. The infrared detection method for weld joint defects based on feature classification according to claim 1, characterized in that: The database in S4 contains thermal characteristic data of different solder joint defect types, and generates a unique information code according to different solder joint defect types.

10. A feature-based infrared detection system for solder joint defects, used to implement the feature-based infrared detection method for solder joint defects according to any one of claims 1-9, characterized in that, The system includes: a data acquisition and processing module, a comparison and filtering module, a feature classification module, and a search and identification module; The acquisition and processing module is used to acquire infrared images of the processed product, locate and determine the solder joint area from the acquired infrared images, filter out the infrared images containing the solder joint area for preprocessing, and generate the input infrared image. The comparison and screening module is used to compare the input infrared image with the complete infrared images in the infrared image complete dictionary to detect whether there are solder joint defects. If there are solder joint defects, first detection information is generated. The feature classification module is used to extract the thermal features of the solder joints from the input infrared image corresponding to the first detection information, input the extracted thermal features of the solder joints into the solder joint detection classifier to determine the defect type, and generate the second detection information based on the defect type. The search identifier module is used to search the database and send an information code based on the second detection information.

Images