Weld defect detection system and method based on target recognition

Abstract

The invention discloses a weld defect detection system and method based on target recognition, and the system comprises an image preprocessing module, a weld defect analysis module and a weld quality analysis module, relates to the technical field of image processing, and solves the problem that data formats and types obtained by different detection methods are different. In order to solve the technical problems of difficulty in effective integration and comprehensive analysis and proneness to erroneous judgment and missing judgment in the prior art, an image preprocessing module carries out graying, image enhancement and denoising processing on a weld seam image, so that the characteristics of weld seam defects can be highlighted, and the definition and contrast of the image are improved; according to the method, more accurate image data is provided for subsequent defect analysis, the image data is combined with historical data, and the relationship among defect types, gray features, defect reasons and the like is established by sorting and analyzing a large amount of historical data, so that the welding seam quality is evaluated more scientifically.

Description

Technical Field

[0001] The present invention relates to the technical field of image processing, and in particular to a weld defect detection system and method based on target recognition. Background Art

[0002] In modern manufacturing, welding technology is widely used in many fields, such as automobile manufacturing, bridge construction, pipeline installation, etc. The quality of the weld is directly related to the safety, reliability and service life of the entire product or structure.

[0003] The patent with publication number CN118429343A discloses a weld quality defect detection system based on a deep convolutional neural network model, which relates to the field of image processing technology. When the system is running, the acquisition module collects image information of the welded parts before and after the test, and the inspection and recording module records the internal state information of the welded parts before and after the test. After statistical analysis, the following is obtained: the welded parts internal state deviation proportional coefficient Pcxs, the collected image information is preprocessed by the analysis module, and then the feature extraction module uses the deep convolutional neural network model for training and extraction to obtain the first data set and the second data set, and the data processing module processes and fits them to obtain: the weld evaluation index Pgz, and then matches it with the pre-examined weld evaluation threshold P of the welded parts to obtain the weld state evaluation strategy of the welded parts.

[0004] Traditional weld inspection methods mainly rely on manual visual inspection and some simple non-destructive testing methods, such as ultrasonic testing and magnetic particle testing. Manual visual inspection is highly subjective and easily affected by factors such as the inspection personnel's experience and fatigue, resulting in inaccurate inspection results and low efficiency.

[0005] Some existing weld defect detection methods are difficult to accurately identify various complex defect types in welds when used, and are prone to misjudgment and missed judgments. At the same time, different detection methods obtain data in different formats and types, making it difficult to conduct effective integration and comprehensive analysis. Summary of the invention

[0006] In view of the shortcomings of the prior art, the present invention provides a weld defect detection system and method based on target recognition, which solves the problem that the data formats and types obtained by different detection methods are different, which makes it difficult to carry out effective integration and comprehensive analysis, and is prone to misjudgment and missed judgment.

[0007] To achieve the above objectives, the present invention is implemented through the following technical solutions: a weld defect detection system based on target recognition, comprising: The image preprocessing module is used to process the weld image transmitted by the weld image acquisition module, obtain the preprocessed image by graying, image enhancement and denoising the weld image, and average the preprocessed image to obtain the segmented image, and transmit the segmented image to the weld defect analysis module; The weld defect analysis module is used to process the acquired segmented image, acquire the grayscale features of the segmented image, determine the defect type according to the grayscale features, and generate a defect image by combining and matching the acquired defect types, and then transmit the defect image to the weld quality analysis module; The weld quality analysis module is used to analyze the acquired defect image, obtain the recognition result by identifying the defect type corresponding to the defect image, generate the specific cause of the defect based on the defect image, and detect the overall quality of the weld according to the generated specific cause of the defect, calculate the weld quality value, and then match it with the quality judgment interval to generate quality detection information, and transmit it to the detection information output module.

[0008] As a further solution of the present invention, it also includes a weld image acquisition module, which is used to acquire the image of the weld and transmit the acquired weld image to the image preprocessing module.

[0009] As a further solution of the present invention, the specific manner in which the image preprocessing module processes the weld image is as follows: The weld image transmitted by the weld image acquisition module is acquired, and the weld image is converted into a grayscale image, and then the obtained grayscale image is subjected to image enhancement processing to obtain an enhanced grayscale image, and the obtained enhanced grayscale image is subjected to denoising processing to obtain a preprocessed image; The preprocessed image is obtained and divided into nine equal parts to obtain a segmented image, which is labeled as i, where i=1, 2, ..., 9. The segmented image is then transmitted to the weld defect analysis module.

[0010] As a further solution of the present invention, the specific manner in which the weld defect analysis module processes the segmented image is as follows: Obtain all segmented images i, and take one group of them as the target object, obtain the grayscale features of the target object, and obtain historical data at the same time. According to the historical data, the grayscale features of different weld defects are sorted to obtain a matching interval, and then the grayscale features of the target object are matched with the matching interval to obtain the defect type of the target object; Similarly, the defect types corresponding to all the segmented images are obtained, and then the segmented images corresponding to the same defect type are screened. At the same time, the screened segmented images are combined and matched to obtain the grayscale features of the segmented images. The segmented images with similar grayscale features are combined according to the original segmentation order to obtain defect images, and the defect images are transmitted to the weld quality analysis module and the detection information output module respectively.

[0011] As a further solution of the present invention, the specific method in which the weld quality analysis module analyzes the defect image is as follows: All defect images are acquired, and the defect types corresponding to the defect images are identified. If there is only one defect type corresponding to the defect image, a single defect signal is generated. Conversely, if there are multiple defect types corresponding to the defect image, multiple defect signals are generated, and both are analyzed separately. Analyze a single defect signal to obtain a defect image, and at the same time obtain the defect features corresponding to the defect image, and the defect features here include defect grayscale value, defect shape and defect texture features, then obtain the defect causes corresponding to all defect features, and at the same time obtain the same defect causes existing in the defect features, and calculate the corresponding proportion values, and then select the defect cause corresponding to the largest proportion value as the standard to generate the specific cause of the defect; The multi-type signals are analyzed in the same way as the single defect signal. For the multi-type signals, different types of defects are analyzed in turn, and the corresponding specific causes of the defects are generated.

[0012] As a further solution of the present invention, the specific manner in which the weld quality analysis module detects the overall quality of the weld according to the specific causes of the generated defects is: Analyze the weld quality based on the specific cause of the defect, obtain the defect length and defect area corresponding to the weld defect, and take their values ​​for calculation. At the same time, sum the obtained defect length and defect area to obtain the defect parameter value, then match the obtained defect specific value with the quality reference table, and obtain the matched assignment; For the assignment corresponding to a single type of signal, the quality of the weld is determined according to the corresponding assignment table, and quality inspection information is generated. For the assignment corresponding to multiple types of signals, the assignments corresponding to all types are calculated, and the sum of the values ​​of all types of assignments is calculated. At the same time, the calculated sum of the values ​​is matched with the assignment table to determine the quality of the weld, and quality inspection information is generated. At the same time, the generated quality inspection information is transmitted to the inspection information output module.

[0013] As a further solution of the present invention, it also includes a detection information output module for displaying the acquired defect image and quality detection information to the corresponding operator.

[0014] A method for detecting weld defects based on target recognition, the method specifically comprising the following steps: Step S1: graying, image enhancement and denoising the obtained weld image to obtain a preprocessed image, and averaging the preprocessed image to obtain a segmented image; Step S2: Processing the obtained segmented image, acquiring the grayscale features of the segmented image, determining the defect type according to the grayscale features, and generating a defect image by performing combination matching based on the obtained defect types; Step S3: Identify the defect type corresponding to the defect image to obtain an identification result, generate a specific cause of the defect based on the defect image, and detect the overall quality of the weld according to the generated specific cause of the defect, calculate the weld quality value, and then match it with the quality judgment interval to generate quality detection information; Step S4: Display the obtained quality inspection information to the corresponding operator.

[0015] Beneficial Effects The present invention provides a weld defect detection system and method based on target recognition. Compared with the prior art, it has the following beneficial effects: The present invention performs grayscale, image enhancement and denoising processing on the weld image through the image preprocessing module, which can highlight the characteristics of weld defects, improve the clarity and contrast of the image, and provide more accurate image data for subsequent defect analysis; The weld defect analysis module uses historical data to establish grayscale feature matching intervals for different weld defects. By accurately matching the grayscale features of the segmented image with the matching intervals, the defect type can be accurately determined. By combining image data with historical data and sorting and analyzing a large amount of historical data, the relationship between defect type and grayscale characteristics, defect causes, etc. is established, providing a more comprehensive basis for weld quality analysis. Different assignment tables are used for quality assessment, and defect length, area and other parameters can be comprehensively calculated. These data are matched with the quality reference table to more scientifically evaluate weld quality. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 This is a block diagram of the system principle of the present invention; Figure 2 It is a diagram of the steps of the present invention. DETAILED DESCRIPTION

[0017] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0018] For example, see Figure 1 The present application provides a weld defect detection system based on target recognition, including: a weld image acquisition module, an image preprocessing module, a weld defect analysis module, a weld quality analysis module and a detection information output module. Figure 1 It can be known that the above functional modules are all electrically connected in one direction.

[0019] The weld image acquisition module is used to acquire the image of the weld and transmit the acquired weld image to the image preprocessing module, and the weld image here is acquired by an optical camera.

[0020] The image preprocessing module is used to process the acquired weld image. The preprocessed image is obtained by graying, image enhancement and denoising the weld image, and the preprocessed image is averaged to obtain a segmented image. The segmented image is transmitted to the weld defect analysis module.

[0021] The weld image transmitted by the weld image acquisition module is obtained, and the weld image is converted into a grayscale image. The grayscale processing can be processed using the weighted average method. For example, the common formula is Gray=0.299R+0.587G+0.114B, where R, G and B are the pixel values ​​of the red, green and blue channels of the color image, respectively, and Gray is the grayscale value. Then, the obtained grayscale image is enhanced to obtain an enhanced grayscale image. The specific image enhancement processing is performed by stretching the contrast of the image. For example, the grayscale value range of the original image is stretched from [a, b] to [c, d]. The linear transformation formula can be used. , where x is the original grayscale value, y is the transformed grayscale value, and the enhanced grayscale image is denoised to obtain a preprocessed image. The denoising method can be used to remove the noise using a filtering method. For example, median filtering is very effective in removing salt and pepper noise. The principle of median filtering is to replace the grayscale value of each pixel with the median of the grayscale values ​​of its neighboring pixels. The preprocessed image is obtained and divided into nine equal parts to obtain a segmented image, which is labeled as i, where i=1, 2, ..., 9. The segmented image is then transmitted to the weld defect analysis module.

[0022] Weld defect analysis module, which is used to process the acquired segmented image by acquiring the grayscale features of the segmented image and determining the defect type based on the grayscale features. At the same time, it generates a defect image by performing combination matching based on the obtained defect type, and then transmits the defect image to the detection information output module.

[0023] Obtain all segmented images i, and take one group of them as the target object, then obtain the grayscale features of the target object, the specific grayscale features represent the grayscale value of the target object, and at the same time obtain historical data, and sort out the grayscale features of different weld defects according to the historical data to obtain a matching interval, and the matching interval here is specifically composed of the grayscale features of the corresponding weld defects, and then match the grayscale features of the target object with the matching interval to obtain the defect type of the target object; For example, an image is selected as the target object, and its grayscale features are extracted using professional image analysis software. For example, after analysis, it is found that the grayscale mean of a suspected defect area in the target object image is 80, and the grayscale variance is 15 (this is the extracted grayscale feature). After sorting out the historical data, it is found that the grayscale mean range corresponding to the hole defect is approximately 60-90, and the grayscale variance range is approximately 10-20; the grayscale mean range corresponding to the slag inclusion defect is approximately 70-100, and the grayscale variance range is approximately 12-25; the grayscale mean range corresponding to the crack defect is approximately 40-70, and the grayscale variance range is approximately 8-15, etc. (these are the grayscale feature matching intervals of different defects). The grayscale feature of the target object is further matched with the matching interval, and it is found that it falls within the grayscale feature matching interval of the pore defect. Therefore, it can be judged that the suspected defect area in the target object image belongs to the pore defect type. The specific identification method is: , the specific identification method is: In this example, the grayscale mean of the suspected defect area in the target object image is 80, and the grayscale variance is 15. The grayscale feature matching intervals of different defects are as follows: Porosity defect: The grayscale mean range is approximately 60-90, and the grayscale variance range is approximately 10-20.

[0024] Slag inclusion defect: The grayscale mean range is approximately 70-100, and the grayscale variance range is approximately 12-25.

[0025] The grayscale mean of the target object is 80, which is within the grayscale mean matching interval (60-90) of the pore defect and the grayscale mean matching interval (70-100) of the slag inclusion defect; but the grayscale variance is 15, which is within the grayscale variance matching interval (10-20) of the pore defect and the grayscale variance matching interval (12-25) of the slag inclusion defect. However, when matching the grayscale variance, although they are matched with the corresponding matching intervals, when calculating the matching difference, the difference between the grayscale variance matching interval (10-20) of the pore defect and the grayscale variance 15 is 5, while the difference between the grayscale variance matching interval of the slag inclusion defect and the grayscale variance 15 is (7-10). Combining the difference between the two, it can be known that when selecting, the matching interval corresponding to the smallest difference is selected for matching, and the corresponding defect cause is generated.

[0026] To sum up, the judgment standard is first to match the grayscale mean, specifically to calculate the difference, and select the grayscale mean with the smallest difference as the standard, and then calculate the grayscale variance, and similarly select the matching interval with the smallest difference.

[0027] Similarly, the defect types corresponding to all the segmented images are obtained, and then the segmented images corresponding to the same defect type are screened, and the screened segmented images are combined and matched, and the specific method of combined matching is: the grayscale features of the segmented images are obtained, and then the segmented images with similar grayscale features are combined. The specific grayscale feature similarity is represented by the grayscale feature difference of the segmented images being within a preset value range, wherein the specific value of the preset value range is set by the operator, and the order of combination is combined according to the original segmentation order to obtain a defect image.

[0028] For example, when inspecting a batch of pipeline welding seams, a series of segmented images were obtained through image analysis. These images were determined to contain different defect types such as pores, slag inclusions, and unfused welds. All segmented images determined to be pore defect types were first screened out. Assume that the grayscale mean of a pore defect segmented image is 75 and the grayscale variance is 18. The operator sets the preset difference range of the grayscale mean to ±5 and the preset difference range of the grayscale variance to ±3 based on experience. Then, among the screened pore defect segmented images, images with grayscale means between 70-80 and grayscale variances between 15-21 will be combined together and arranged and combined according to the order in which these images were originally in the inspection sequence.

[0029] The detection information output module is used to display the acquired defect image to the corresponding operator.

[0030] Embodiment 2: This embodiment is implemented on the basis of Embodiment 1, and the difference from Embodiment 1 is as follows: The weld defect analysis module transmits the generated defect image information to the weld quality analysis module, and the weld is inspected by the weld quality analysis module.

[0031] Weld quality analysis module, which is used to analyze the acquired defect image, obtain the recognition result by identifying the defect type corresponding to the defect image, generate the specific cause of the defect based on the defect image, and detect the overall quality of the weld according to the generated specific cause of the defect, calculate the weld quality value, and then match it with the quality judgment interval to generate quality detection information, and transmit it to the detection information output module.

[0032] All defect images are acquired, and the defect types corresponding to the defect images are identified. If there is only one defect type corresponding to the defect image, a single defect signal is generated. Conversely, if there are multiple defect types corresponding to the defect image, multiple defect signals are generated, and both are analyzed separately. For example, after inspecting a welding part of a steel structure, all defect images obtained are judged to be porosity defects. At this time, a single defect signal will be generated. For example, some defect images show slag inclusion defects, while others show unfusion defects. In this case, multiple types of defect signals will be generated, which means that there are multiple quality problems in the welding area.

[0033] A single defect signal is analyzed to obtain a defect image, and the defect features corresponding to the defect image are obtained at the same time. The defect features here include defect grayscale value, defect shape and defect texture features. Then, the defect causes corresponding to all defect features are obtained, and the defect causes obtained here are specifically expressed as defect causes corresponding to defect grayscale value, defect shape and defect texture features respectively. At the same time, the same defect causes existing in the defect features are obtained, and the corresponding proportion values ​​are calculated. For example, the defect causes obtained after analysis include welding temperature, welding material quality, welding molten pool abnormality and impurity mixing. The welding temperature exists three times in the corresponding defect feature, and the corresponding proportion is three quarters, and the welding material quality exists once, and the corresponding proportion is one quarter. Similarly, the proportion values ​​of the remaining specific causes are determined, and then the defect cause corresponding to the largest proportion value is selected as the standard to generate the specific cause of the defect. Following the above example, in the above example, assuming that the proportion value corresponding to the welding temperature is the largest, the welding temperature is used as the standard as the specific cause of the defect.

[0034] Analyze multiple types of signals in the same way as single defect signals. For multiple types of signals, analyze different types of defects in turn and generate the corresponding specific causes of the defects. The weld quality is analyzed based on the specific causes of the defects, the defect length and defect area corresponding to the weld defects are obtained, and the values ​​are taken for calculation. At the same time, the obtained defect length and defect area are summed to obtain the defect parameter value, and then the obtained specific defect value is matched with the quality reference table. The quality reference table here is set by the operator according to historical data, specifically composed of specific defect values, and the matched assignment is obtained; For the assignment corresponding to a single type of signal, the quality of the weld is determined according to the corresponding assignment table, and quality inspection information is generated. The assignment table is set by the operator. For the assignment corresponding to multiple types of signals, the assignments corresponding to all types are calculated, and the sum of the values ​​of all types of assignments is calculated. At the same time, the calculated sum of the values ​​is matched with the assignment table. The assignment table here is different from the assignment table corresponding to a single type of signal. It is intended to comprehensively consider the impact of multiple defects on the quality of the weld, determine the quality of the weld, and generate quality inspection information. At the same time, the generated quality inspection information is transmitted to the inspection information output module.

[0035] For example, when inspecting the weld of a bridge steel structure, a defect was found in one weld. The defect length was measured to be 5 mm and the defect area was 3 square millimeters. The sum of the two gave a defect parameter value of 8. In the quality reference table, if the defect parameter value is less than 10, the corresponding value is 80 (indicating a higher quality level), then the value assigned to the weld here is 80. For example, if the value is 80 or above, the corresponding quality level in the value assignment table is "poor"; When multiple types of signals are detected, such as the presence of slag inclusion and unfusion defect signals, the corresponding values ​​for each type of defect are calculated separately. Assume that the slag inclusion defect is assigned a value of 60 and the unfusion defect is assigned a value of 50. Then calculate the sum of all types of assignments to be 110. Then match this sum of values ​​with the assignment table specifically set for multiple types of signals. This assignment table has a different structure and standard from the assignment table for a single type of signal, and is designed to comprehensively consider the impact of multiple defects on weld quality. For example, in the multi-type assignment table, the quality level corresponding to the sum of values ​​between 100-120 is "medium", so it can be determined that the weld quality is "medium".

[0036] The detection information output module is used to display the acquired quality detection information to the corresponding operator.

[0037] Embodiment 3, as the third implementation of the present invention, focuses on combining the implementation processes of Embodiment 1 and Embodiment 2.

[0038] Example 4, please refer to Figure 2 The present application provides a weld defect detection method based on target recognition, which specifically includes the following steps: Step S1: graying, image enhancement and denoising the obtained weld image to obtain a preprocessed image, and averaging the preprocessed image to obtain a segmented image, and the specific processing method is the same as the processing process of the image preprocessing module in Example 1; Step S2: Processing the obtained segmented image, acquiring the grayscale features of the segmented image, determining the defect type according to the grayscale features, and generating a defect image by combining and matching the obtained defect types. The processing method here is the same as the processing process of the weld defect analysis module in Example 1. Step S3: Identify the defect type corresponding to the defect image to obtain an identification result, generate a specific defect cause based on the defect image, and detect the overall quality of the weld according to the generated specific defect cause, calculate the weld quality value, and then match it with the quality judgment interval to generate quality detection information, and the processing method here is the same as the processing process of the weld defect analysis module in Example 2; Step S4: Display the obtained quality inspection information to the corresponding operator.

[0039] Meanwhile, the contents not described in detail in this specification belong to the prior art known to those skilled in the art.

[0040] The above embodiments are only used to illustrate the technical method of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical method of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical method of the present invention.

Claims

1. A weld defect detection system based on target recognition, characterized in that: include: The image preprocessing module is used to process the weld image transmitted by the weld image acquisition module, obtain the preprocessed image by graying, image enhancement and denoising the weld image, and average the preprocessed image to obtain the segmented image, and transmit the segmented image to the weld defect analysis module; The weld defect analysis module is used to process the acquired segmented image, acquire the grayscale features of the segmented image, determine the defect type according to the grayscale features, and generate a defect image by combining and matching the acquired defect types, and then transmit the defect image to the weld quality analysis module; The weld quality analysis module is used to analyze the acquired defect image, obtain the recognition result by identifying the defect type corresponding to the defect image, generate the specific cause of the defect based on the defect image, and detect the overall quality of the weld according to the generated specific cause of the defect, calculate the weld quality value, and then match it with the quality judgment interval to generate quality detection information, and transmit it to the detection information output module.

2. A weld defect detection system based on target recognition according to claim 1, characterized in that: It also includes a weld image acquisition module, which is used to acquire the image of the weld and transmit the acquired weld image to the image preprocessing module.

3. A weld defect detection system based on target recognition according to claim 1, characterized in that: The specific method in which the image preprocessing module processes the weld image is as follows: The weld image transmitted by the weld image acquisition module is acquired, and the weld image is converted into a grayscale image, and then the obtained grayscale image is subjected to image enhancement processing to obtain an enhanced grayscale image, and the obtained enhanced grayscale image is subjected to denoising processing to obtain a preprocessed image; The preprocessed image is obtained and divided into nine equal parts to obtain a segmented image, which is labeled as i, where i=1, 2, ..., 9. The segmented image is then transmitted to the weld defect analysis module.

4. A weld defect detection system based on target recognition according to claim 1, characterized in that: The specific method in which the weld defect analysis module processes the segmented image is as follows: Obtain all segmented images i, and take one group of them as the target object, obtain the grayscale features of the target object, and obtain historical data at the same time. According to the historical data, the grayscale features of different weld defects are sorted to obtain a matching interval, and then the grayscale features of the target object are matched with the matching interval to obtain the defect type of the target object; Similarly, the defect types corresponding to all the segmented images are obtained, and then the segmented images corresponding to the same defect type are screened. At the same time, the screened segmented images are combined and matched to obtain the grayscale features of the segmented images. The segmented images with similar grayscale features are combined according to the original segmentation order to obtain defect images, and the defect images are transmitted to the weld quality analysis module and the detection information output module respectively.

5. The weld defect detection system based on target recognition according to claim 1 is characterized in that: The specific method in which the weld quality analysis module analyzes the defect image is as follows: All defect images are acquired, and the defect types corresponding to the defect images are identified. If there is only one defect type corresponding to the defect image, a single defect signal is generated. Conversely, if there are multiple defect types corresponding to the defect image, multiple defect signals are generated, and both are analyzed separately. Analyze a single defect signal to obtain a defect image, and at the same time obtain the defect features corresponding to the defect image, and the defect features here include defect grayscale value, defect shape and defect texture features, then obtain the defect causes corresponding to all defect features, and at the same time obtain the same defect causes existing in the defect features, and calculate the corresponding proportion values, and then select the defect cause corresponding to the largest proportion value as the standard to generate the specific cause of the defect; The multi-type signals are analyzed in the same way as the single defect signal. For the multi-type signals, different types of defects are analyzed in turn, and the corresponding specific causes of the defects are generated.

6. A weld defect detection system based on target recognition according to claim 1, characterized in that: The specific method in which the weld quality analysis module detects the overall quality of the weld according to the specific causes of the generated defects is: Analyze the weld quality based on the specific cause of the defect, obtain the defect length and defect area corresponding to the weld defect, and take their values ​​for calculation. At the same time, sum the obtained defect length and defect area to obtain the defect parameter value, then match the obtained defect specific value with the quality reference table, and obtain the matched assignment; For the assignment corresponding to a single type of signal, the quality of the weld is determined according to the corresponding assignment table, and quality inspection information is generated. For the assignment corresponding to multiple types of signals, the assignments corresponding to all types are calculated, and the sum of the values ​​of all types of assignments is calculated. At the same time, the calculated sum of the values ​​is matched with the assignment table to determine the quality of the weld, and quality inspection information is generated. At the same time, the generated quality inspection information is transmitted to the inspection information output module.

7. A weld defect detection system based on target recognition according to claim 4, characterized in that: It also includes a detection information output module for displaying the acquired defect images and quality detection information to the corresponding operators.

8. A weld defect detection method based on target recognition, applied to the weld defect detection system according to any one of claims 1 to 7, characterized in that: The method specifically comprises the following steps: Step S1: graying, image enhancement and denoising the obtained weld image to obtain a preprocessed image, and averaging the preprocessed image to obtain a segmented image; Step S2: Processing the obtained segmented image, acquiring the grayscale features of the segmented image, determining the defect type according to the grayscale features, and generating a defect image by performing combination matching based on the obtained defect types; Step S3: Identify the defect type corresponding to the defect image to obtain an identification result, generate a specific cause of the defect based on the defect image, and detect the overall quality of the weld according to the generated specific cause of the defect, calculate the weld quality value, and then match it with the quality judgment interval to generate quality detection information; Step S4: Display the obtained quality inspection information to the corresponding operator.

Images