Defect identification method, system and equipment for automatic coil detection machine

Abstract

The invention provides a defect identification method, system and equipment of an automatic coil detector, and relates to the technical field of image processing, the method comprises the following steps: executing image acquisition when a coil reaches a target detection area; performing distortion and illumination correction on the detection image set, performing multispectral feature fusion on the corrected image set, inputting an enhanced image into a window identification channel to perform image traversal scanning of the enhanced image, and calculating texture energy and a direction consistency index in each sliding window; performing multi-scale image anomaly comparison by using the calibration position image and the enhanced image corresponding to the coordinate index; and performing morphological aggregation and boundary continuity analysis on the pixel points with the abnormal value identifiers, extracting geometric morphological parameters, spectral characteristic parameters and texture abnormal parameters, and outputting defect masks. According to the invention, the technical problem of low defect identification precision of the coil detection machine in the prior art can be solved, and the technical effect of improving the defect identification precision is achieved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of image processing, and particularly relates to a defect identification method, system and device of a coil automatic detection machine. BACKGROUND

[0002] In modern industrial production, as an important component of electrical equipment and precision machinery, the quality of the coil directly affects the performance and service life of the equipment, so defect detection and quality control of the coil become a key link.

[0003] At present, existing coil automatic detection technology mainly relies on single optical imaging or simple infrared detection means, and the surface of the coil is identified for defects by manual or semi-automatic means. This kind of technology usually uses a single spectral image acquisition device, and the collected images are prone to distortion and uneven brightness under the conditions of light, imaging angle and environmental interference, thereby affecting the accurate extraction of defect features.

[0004] In summary, in the prior art, due to the dependence on single spectral imaging, the lack of multi-feature and multi-scale image fusion analysis, and the low defect positioning precision, it is difficult to accurately identify and quantify the small defects and local abnormalities on the surface of the coil, which further affects the real-time monitoring of the production process, the defect tracing ability, and the stability and reliability of the quality of the coil product. SUMMARY

[0005] The purpose of the present application is to provide a defect identification method, system and device of a coil automatic detection machine, to solve the technical problems in the prior art that due to the dependence on single spectral imaging, the lack of multi-feature and multi-scale image fusion analysis, and the low defect positioning precision, it is difficult to accurately identify and quantify the small defects and local abnormalities on the surface of the coil, which further affects the real-time monitoring of the production process, the defect tracing ability, and the stability and reliability of the quality of the coil product.

[0006] In view of the above problems, the present application provides a defect identification method, system and device of a coil automatic detection machine.

[0007] In a first aspect, the application provides a defect identification method of a coil automatic detection machine, which is realized by a defect identification system of a coil automatic detection machine, and includes the following steps: when a coil reaches a target detection area, a detection instruction of the coil automatic detection machine is triggered; after an annular adjustable light source is started according to the detection instruction, a visible light camera and an infrared camera arranged directly above the target detection area are controlled to perform image acquisition, and a detection image set is established; after distortion and illumination correction are performed on the detection image set, the corrected image set is subjected to multi-spectrum feature fusion, an enhanced image is constructed, the enhanced image is input into a window recognition channel, image traversal scanning of the enhanced image is performed by using a sliding window, texture energy and direction consistency indexes in each sliding window are calculated, and a coordinate index is established; a calibration position image is called according to the coordinate index, multi-scale image anomaly comparison is performed on the calibration position image and the enhanced image corresponding to the coordinate index, and a pixel point with an anomaly value identifier is established; after the pixel point with the anomaly value identifier is subjected to morphological aggregation and boundary continuity analysis, geometric morphological parameters, spectral feature parameters and texture anomaly parameters are extracted, and a defect mask is output.

[0008] Preferably, the defect identification method of the coil automatic detection machine further includes the following steps: a three-dimensional calibration model of the target detection area is called from a calibration database of the detection machine, and radial distortion and tangential distortion joint correction is performed on the visible light image and the infrared image respectively by using the three-dimensional calibration model; environmental spectrum monitoring data of the target detection area are acquired, a light distribution surface model is constructed based on the environmental spectrum monitoring data, and illumination correction is performed on the visible light image and the infrared image respectively after distortion correction by using the light distribution surface model under block brightness equalization; the following steps are performed: the visible light image or the infrared image is divided into N grid blocks, and a stable brightness statistic is calculated for each grid block; after a target brightness constant is configured according to the stable brightness statistic, a gain map of each pixel is calculated, and multiplicative correction of the original image is performed to complete the illumination correction.

[0009] Preferably, the defect identification method of the coil automatic detection machine further includes the following steps: after the visible light image and the infrared image after illumination correction are subjected to pixel-level spatial geometric registration, layered feature decomposition is performed, low-frequency layer features, medium-frequency layer features and high-frequency layer features are extracted, and adaptive weight fusion is performed based on a definition score of the multi-layer features and a defect prior probability; and the enhanced image is constructed by using the adaptive weight fusion result.

[0010] Preferably, the defect identification method of the coil automatic detection machine further comprises: calling a texture extraction layer of the window identification channel, performing enhanced image inner texture complexity evaluation, and establishing a texture complexity evaluation identifier; establishing a window shape library of a sliding window, the window shape library comprising a square window and a rectangular window; after shape matching of the window shape library is performed using the texture complexity evaluation identifier, a position window shape constraint is generated; after a multi-scale position window is configured using the position window shape constraint, texture energy descriptor calculation is performed, a texture energy distribution is established, and a coordinate index is established according to the texture energy distribution and a direction consistency index.

[0011] Preferably, the defect identification method of the coil automatic detection machine further comprises: performing gradient direction clustering on each multi-scale position window, and calculating a main direction concentration degree; taking the main direction concentration degree as a direction consistency index, and establishing a coordinate index according to the texture energy distribution and the direction consistency index.

[0012] Preferably, the defect identification method of the coil automatic detection machine further comprises: calling a multi-scale calibration position image in a calibration library according to the coordinate index; after performing set correction and spectral normalization processing on the multi-scale calibration position image, performing image registration at different scale layers, and performing abnormal comparison based on spectral difference, texture difference and set difference, an abnormal value identifier is established.

[0013] Preferably, the defect identification method of the coil automatic detection machine further comprises: performing spatial aggregation of abnormal pixels using morphological-based connected region analysis, and merging adjacent similar abnormal pixel points, the connected region analysis comprising expansion region analysis, corrosion region analysis and open-close operation region analysis; calculating a boundary curvature and a concave-convex index for each spatial aggregation result, and generating a boundary continuity analysis result; and extracting a geometric morphological parameter, a spectral feature parameter and a texture abnormality parameter using the spatial aggregation result and the boundary continuity analysis result.

[0014] Preferably, the defect identification method of the coil automatic detection machine further comprises: obtaining a parameter mapping of coil manufacturing parameters and a defect set; after obtaining a defect mask, performing associated matching of manufacturing parameters according to the defect mask and the parameter mapping, and generating an associated record; generating production feedback according to the associated record, and using the production feedback for coil production management.

[0015] In a second aspect, the application further provides a defect identification system of a coil automatic detection machine, which is used to execute the defect identification method of the coil automatic detection machine as described in the first aspect, and comprises: a detection instruction activation module, which is used to trigger the detection instruction of the coil automatic detection machine when the coil reaches the target detection area; an image acquisition module, which is used to control the visible light camera and the infrared camera arranged directly above the target detection area to perform image acquisition and establish a detection image set after the ring-shaped adjustable light source is started according to the detection instruction; an image traversal scanning module, which is used to perform distortion and illumination correction on the detection image set, perform multi-spectrum feature fusion on the corrected image set, construct an enhanced image, input the enhanced image into a window recognition channel, perform image traversal scanning of the enhanced image by using a sliding window, calculate the texture energy and the direction consistency index in each sliding window, and establish a coordinate index; an image anomaly comparison module, which is used to call a calibration position image according to the coordinate index, perform multi-scale image anomaly comparison on the enhanced image corresponding to the calibration position image and the coordinate index, and establish a pixel point with an abnormal value identifier; and a defect mask output module, which is used to perform morphological aggregation and boundary continuity analysis on the pixel point with the abnormal value identifier, extract geometric morphological parameters, spectral feature parameters and texture anomaly parameters, and output a defect mask.

[0016] In a third aspect, the application further provides an electronic device, which comprises: at least one processor; and a memory connected with the at least one processor in communication; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute the steps of the defect identification method of the coil automatic detection machine according to any one of the first aspect.

[0017] The technical solutions provided in the application have at least the following technical effects or advantages: the technical target of high-precision, full-spectrum and multi-scale automatic identification and positioning of the surface defects of the coil is achieved, the technical effects of improving the defect detection accuracy, enhancing the defect quantitative analysis capability, realizing the closed-loop feedback of production parameters and optimizing the coil production management are achieved.

[0018] The above description is only a summary of the technical solutions of the application. In order to enable the technical means of the application to be more clearly understood, and to be implemented according to the content of the description, and in order to enable the above and other purposes, characteristics and advantages of the application to be more apparent and easy to understand, the specific implementation manner of the application is described 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 application, nor is it used to limit the scope of the application. Other features of the application will become apparent from the following description. BRIEF DESCRIPTION OF DRAWINGS

[0019] In order to more clearly illustrate the technical solutions in the application or the prior art, the accompanying drawings needed to be used in the description of the embodiments or the prior art will be briefly introduced. Obviously, the accompanying drawings in the following description are only exemplary and, for those skilled in the art, other drawings can be obtained without creative effort on the basis of the provided drawings.

[0020] Figure 1 A defect identification method flow chart of a coil automatic detection machine provided for an embodiment of the application.

[0021] Figure 2 A defect identification system structure schematic diagram of a coil automatic detection machine provided for an embodiment of the application.

[0022] Figure 3 A structure schematic diagram of an exemplary electronic device of the application.

[0023] Legend of the accompanying drawings: detection instruction activation module 1, image acquisition module 2, image traversal scanning module 3, image anomaly comparison module 4, defect mask output module 5, bus 300, receiver 301, processor 302, transmitter 303, memory 304, bus interface 305. DETAILED DESCRIPTION

[0024] The application provides a defect identification method, system and device of a coil automatic detection machine, which solves the technical problem in the prior art that due to the dependence on single spectral imaging, the lack of multi-feature and multi-scale image fusion analysis, and the low defect positioning precision, it is difficult to accurately identify and quantify the small defects and local anomalies on the surface of the coil, which further affects the real-time monitoring of the production process, the defect tracing capability, and the stability and reliability of the quality of the coil product. The technical goal of high-precision, full-spectrum, multi-scale automatic identification and positioning of the defects on the surface of the coil is achieved, and the technical effects of improving the defect detection accuracy, enhancing the defect quantitative analysis capability, realizing the closed-loop feedback of the production parameters, and optimizing the coil production management are achieved.

[0025] The technical solutions in the application will be described clearly and completely below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the application, rather than all the embodiments of the application. It should be understood that the application is not limited by the exemplary embodiments described herein. Based on the embodiments of the application, all other embodiments obtained by those skilled in the art without creative effort belong to the scope of protection of the application. In addition, it should be noted that, for convenience of description, only parts related to the application are shown in the drawings, not all.

[0026] Embodiment one, please refer to the accompanying Figure 1The application provides a defect identification method of a coil automatic detection machine, which is applied to a defect identification system of a coil automatic detection machine and specifically includes the following steps. S1: When the coil reaches the target detection area, trigger the detection instruction of the coil automatic detection machine.

[0027] Specifically, the target detection area is a working space for performing detection. When the coil enters the pre-set detection position, the coil has reached the target detection area, which is detected by a position sensor or other triggering device. Once the triggering signal is identified, the detection instruction is immediately issued to coordinate the working timing of each detection device.

[0028] S2: After starting the annular adjustable light source according to the detection instruction, control the visible light camera and the infrared camera arranged directly above the target detection area to perform image acquisition and establish a detection image set.

[0029] Specifically, the detection instruction controls the annular adjustable light source to start. The annular adjustable light source is annularly distributed, can uniformly irradiate the surface of the coil from multiple angles, and can adjust the brightness or spectral distribution to adapt to different material reflectivity or environmental light changes, so as to ensure that the acquired images are uniformly illuminated.

[0030] At the same time of starting the annular adjustable light source, drive the two types of cameras installed directly above the target detection area to work synchronously. The visible light camera acquires images in the visible band of the human eye, which can reflect the color, shape and obvious defects of the coil surface. The infrared camera acquires thermal radiation information, which can find temperature differences or internal abnormalities that are difficult to identify by the naked eye. The images acquired by the visible light camera and the infrared camera together constitute the detection image set, which is used for subsequent defect identification.

[0031] S3: After distortion and illumination correction of the detection image set, perform multi-spectral feature fusion on the corrected image set, construct an enhanced image, input the enhanced image into a window recognition channel, use a sliding window to perform image traversal scanning of the enhanced image, calculate the texture energy and direction consistency index in each sliding window, and establish a coordinate index.

[0032] Specifically, the distortion and illumination correction of the detection image set eliminates the deformation of the detection image set caused by the lens or imaging angle. For example, radial distortion will cause barrel or pillow deformation on the edge of the image, and tangential distortion will cause uneven distortion of the image. At the same time, the brightness difference caused by uneven illumination is corrected, so that the detection image set is more true and accurate in overall brightness and geometric shape. The multi-spectral feature fusion of the corrected image set aligns and extracts features of image information in different spectral bands, such as visible light and infrared light, at the pixel level, to retain more details and spectral differences, construct an enhanced image, and generate an image containing higher contrast, clearer details and more complete texture, so that subsequent analysis is easier to find defects.

[0033] The enhanced image is input into the window recognition channel, and then sent to the image analysis module for region-by-region scanning and texture feature recognition. A sliding window is used to perform image traversal scanning of the enhanced image; that is, a window of fixed or varying size is gradually slid across the image, analyzing image information region by region. The strength of the texture in a region is measured by statistically analyzing the intensity of pixel grayscale changes, quantifying the stability of the texture direction, recording the position of each window and its texture energy and direction consistency values, and establishing a coordinate index to provide reference data for subsequent precise defect location.

[0034] S4: Call the calibration location image according to the coordinate index, and use the calibration location image and the enhanced image corresponding to the coordinate index to perform multi-scale image anomaly comparison and establish pixel points for anomaly identification.

[0035] Specifically, the calibration location image is retrieved based on the coordinate index. This involves using index information containing the coordinates of specific locations to retrieve the corresponding reference image from the calibration library. The calibration location image serves as the comparison benchmark during detection. Multi-scale image anomaly comparison is performed using the calibration location image and the enhanced image corresponding to the coordinate index. This involves comparing the reference image and the detected enhanced image layer by layer at multiple resolutions or observation scales, analyzing everything from overall contours to subtle textures to comprehensively detect anomalies. Pixels with differences exceeding a set threshold are marked, establishing anomaly markers.

[0036] S5: After performing morphological aggregation and boundary continuity analysis on the pixels with outlier identifiers, extract geometric morphological parameters, spectral feature parameters and texture anomaly parameters, and output the defect mask.

[0037] Specifically, after performing morphological aggregation and boundary continuity analysis on pixels marked as outliers, the individual pixels identified as anomalous are combined according to spatial relationships and adjacency to form more complete anomalous regions. Simultaneously, the continuity, curvature, and concavity / convexity of the region boundaries are analyzed to ensure smooth and complete boundaries and to eliminate isolated noise points. Morphological aggregation can be achieved through methods such as dilation, erosion, and opening / closing operations, while boundary continuity analysis evaluates the shape integrity of the anomalous region by calculating boundary curvature and concavity / convexity indices.

[0038] Quantitative indexes describing defect shape, color spectrum characteristics and texture changes are obtained from the abnormal area after polymerization, and geometric parameters, spectral feature parameters and texture abnormality parameters are extracted. The geometric parameters include area, perimeter and boundary curvature, the spectral feature parameters represent the difference between the abnormal area and the surrounding area in color or infrared waveband, and the texture abnormality parameters measure the texture intensity or direction consistency of the area. A binary image with the same size as the original image is generated, and a defect mask is output. The pixels of the abnormal area are marked as 1 or highlighted, and the pixels of the normal area are marked as 0 or remain the original color.

[0039] Further, the present application also includes: calling a three-dimensional calibration model of the target detection area through the calibration database of the detection machine, performing radial distortion and tangential distortion joint correction on the visible light image and the infrared image respectively by using the three-dimensional calibration model; obtaining environmental spectrum monitoring data of the target detection area, constructing an illumination distribution surface model based on the environmental spectrum monitoring data, and performing illumination correction on the visible light image and the infrared image respectively after distortion correction by using the illumination distribution surface model under block brightness equalization; and performing the following: dividing the visible light image or the infrared image into N grid blocks, calculating a stable brightness statistic for each grid block; after configuring a target brightness constant according to the stable brightness statistic, calculating a gain map for each pixel, and performing multiplicative correction on the original image to complete the illumination correction.

[0040] Specifically, the calibration database of the detection machine is a set of pre-stored device calibration information library, containing geometric and optical parameters of various detection areas. The three-dimensional calibration model of the target detection area is called through the calibration database of the detection machine, and the three-dimensional calibration model corresponding to the target detection area is read from the calibration database, so that the spatial relationship between the camera, the light source and the detection object can be accurately described.

[0041] The three-dimensional calibration model is used to perform radial distortion and tangential distortion joint correction on the visible light image and the infrared image, to correct the image bending deformation caused by the lens characteristics. The radial distortion refers to the magnification or compression phenomenon occurring from the image center to the outside, and the tangential distortion refers to the tilt deformation caused by the fact that the lens is not completely parallel to the imaging plane. Joint correction can ensure the geometric accuracy of the image.

[0042] The environmental spectrum monitoring data of the target detection area is obtained, which is the light intensity and spectral distribution information of the detection area obtained in real time by the spectral sensor, and can reflect the comprehensive influence of the light source and the environment light. The illumination distribution surface model is constructed based on the environmental spectrum monitoring data, which is used to describe the strength change of the light at different positions. The illumination correction is performed on the visible light image and the infrared image after distortion correction by using the illumination distribution surface model under block brightness equalization, which can compensate for the problem of local area brightness unevenness, and improve the image contrast and detail performance.

[0043] Wherein, the visible light image or infrared image is divided into N grid blocks for analyzing the brightness characteristics in a local range, and the size of N is set according to the image resolution and detection accuracy requirements. The stable brightness statistics of each grid block is calculated, and the relatively stable brightness mean or median under the illumination fluctuation is counted to reduce the interference of transient light changes on the results.

[0044] The target brightness constant is configured according to the stable brightness statistics, and a desired brightness reference is set for calculating the gain map of each pixel to determine the proportion of brightness that each pixel needs to increase or decrease. Then the multiplicative correction of the original image is performed, and the brightness of each pixel is multiplied by the corresponding gain coefficient, so as to unify the brightness distribution and realize the illumination correction. Table 1 is a target detection area image correction data table.

[0045] Table 1: Target detection area image correction data table

[0046] Further, the present application also includes: after the visible light image and the infrared image after the illumination correction are pixel-level spatial geometrically registered, performing hierarchical feature decomposition, extracting low-frequency layer features, medium-frequency layer features, and high-frequency layer features, and performing adaptive weight fusion based on the clarity score of the multi-layer features and the defect prior probability; and constructing an enhanced image using the adaptive weight fusion result.

[0047] Specifically, the visible light image and the infrared image after the illumination correction are pixel-level spatial geometrically registered, and the visible light image and the infrared image are accurately aligned at each pixel position, so that the pixels at the same physical position correspond to each other in the visible light image and the infrared image. Wherein, the spatial geometric registration needs to consider the imaging angle difference, lens distortion and position offset and other factors, so as to ensure the accuracy of subsequent feature fusion.

[0048] After registration is completed, hierarchical feature decomposition is performed to decompose the registered image information into different frequency levels, including low-frequency layer features, medium-frequency layer features, and high-frequency layer features, wherein the low-frequency layer features contain overall shape and brightness distribution, the medium-frequency layer features contain local structure and contour information, and the high-frequency layer features contain detail and texture information, so that different levels of features can be analyzed and processed respectively.

[0049] After the low-frequency layer features, medium-frequency layer features and high-frequency layer features are extracted, adaptive weight fusion is performed according to the clarity score of the multi-layer features and the defect prior probability, the clarity score is used to measure the distinguishability of the multi-layer features, the defect prior probability is based on historical samples or detection model to predict the probability that the multi-layer features may contain defect information, and the adaptive weight fusion is to dynamically allocate the weight proportion of each layer feature in the final result according to the index, so that the defect information is more prominent.

[0050] The adaptive weight fusion result is used to construct an enhanced image, and the fused multi-layer features are recombined into a complete image, thereby retaining the key structure and spectral information in the visible light image and the infrared image, and having higher contrast and detail performance in the defect area, so as to provide better data input for subsequent detection.

[0051] Further, the application also includes: calling the texture extraction layer of the window recognition channel, performing texture complexity evaluation in the enhanced image, and establishing a texture complexity evaluation identifier; establishing a window shape library of the sliding window, the window shape library including square windows and rectangular windows, generating position window shape constraints after shape matching of the window shape library using the texture complexity evaluation identifier; performing texture energy descriptor calculation after configuring multi-scale position windows using the position window shape constraints, establishing a texture energy distribution, and establishing a coordinate index according to the texture energy distribution and the direction consistency index.

[0052] Specifically, the texture extraction layer of the window recognition channel is called, a processing module for analyzing image texture information is enabled, and the rules of gray or color changes in the enhanced image are identified. Then, texture complexity evaluation in the enhanced image is performed, the degree of detail change of the texture in the enhanced image is calculated to measure its complexity, for example, by quantifying the complexity through statistics of local contrast, direction change and frequency component, the calculation result is converted into a label or a numerical value available for subsequent steps, and a texture complexity evaluation identifier is established.

[0053] A window shape library of the sliding window is established for scanning the enhanced image, including different shapes such as square windows and rectangular windows to adapt to different defect shapes and sizes. Shape matching of the window shape library using the texture complexity evaluation identifier is to select the most suitable window shape according to the texture features of the image area, for example, a rectangular window can be selected for a region with regular texture direction, and a square window can be selected for a region with irregular texture direction, thereby specifying the window shape used for each scanning position and generating position window shape constraints.

[0054] After configuring multi-scale position windows using the position window shape constraints, the size of the window is changed on the basis of the fixed shape to cover defect areas of different sizes. Texture energy descriptor calculation is performed to calculate texture energy feature quantities in each window, and the sum of squares of pixel gray changes is counted, etc., all texture energy results of the positions are summarized into a distribution diagram to measure the texture intensity, a texture energy distribution is established to reflect the texture characteristics of different regions of the whole image. According to the texture energy distribution and the direction consistency index, a coordinate index is established, the important detection positions in the enhanced image are recorded in combination with the stability of the texture intensity and the texture direction, to facilitate subsequent positioning and analysis.

[0055] Further, the application further comprises: performing gradient direction clustering on each multi-scale position window, and calculating a main direction concentration; taking the main direction concentration as a direction consistency index, and establishing a coordinate index according to the texture energy distribution and the direction consistency index.

[0056] Specifically, gradient direction clustering is performed on each multi-scale position window, the direction information of pixel gray level change in different sizes of multi-scale position windows is analyzed, gradient vectors of similar directions are classified into a class, so as to identify the main arrangement direction of the texture or edge. The proportion of the class with the highest frequency in all classes is counted to obtain the main direction concentration. The higher the proportion of the main direction concentration, the more uniform the texture direction of the region.

[0057] The main direction concentration is taken as a direction consistency index, which is used to measure the stability of the texture direction. The higher the main direction concentration, the better the direction consistency, and the lower the concentration, the more chaotic the texture direction. According to the texture energy distribution and the direction consistency index, a coordinate index is established, that is, the texture intensity and the direction stability are combined to mark each position in the enhanced image, thereby generating an index containing key position coordinates, which is used for subsequent defect comparison and positioning.

[0058] Further, the application further comprises: calling multi-scale calibration position images in the calibration library according to the coordinate index; performing set correction and spectral normalization processing on the multi-scale calibration position images, performing image registration after different scale layers, and performing abnormal comparison based on spectral difference, texture difference and set difference to establish an abnormal value identification.

[0059] Specifically, multi-scale calibration position images in the calibration library are called according to the coordinate index, that is, the coordinate index containing position and feature information is used to call out images of multiple resolutions or scales corresponding to the position from the standard reference images stored in the calibration library. The multi-scale calibration position images can provide different detail levels from the whole to the local, which is convenient for subsequent comparison.

[0060] Set correction and spectral normalization processing are performed on the multi-scale calibration position images to uniformly correct the geometry and brightness of the reference images of different scales, so that the multi-scale calibration position images are consistent with the current detection images in shape and color spectrum. Set correction refers to the uniformity of scale, rotation and position relationship between multiple images, and spectral normalization processing eliminates color and brightness deviations caused by imaging equipment or environmental light differences.

[0061] After performing image registration at different scale layers, the reference image and the detection image are accurately aligned at the pixel level at each scale layer to ensure that each pixel position in subsequent difference comparison can be correctly corresponded. Then, based on spectral differences, texture differences, and set differences, the abnormality is compared, the differences in color spectrum, changes in surface texture, and deviations in overall morphology or structure are compared respectively, so that the area not meeting the reference standard is found, and an abnormal value identifier is established.

[0062] Further, the application further comprises: performing spatial aggregation of abnormal pixels using morphological-based connected region analysis, merging adjacent similar abnormal pixels, the connected region analysis comprising dilation region analysis, erosion region analysis, and open-close operation region analysis; calculating boundary curvature and concave-convex index for each spatial aggregation result to generate a boundary continuity analysis result; and extracting geometric morphological parameters, spectral feature parameters, and texture abnormality parameters using the spatial aggregation result and the boundary continuity analysis result.

[0063] Specifically, the spatial aggregation of abnormal pixels is performed using morphological-based connected region analysis, that is, the pixels marked as abnormal in the detection are grouped according to the spatial connection relationship, so that pixels adjacent to each other and similar in characteristics are merged into larger abnormal regions. The dilation region analysis in the connected region analysis refers to adding pixels on the pixel boundary, so that adjacent but not directly connected abnormal points are connected together; the erosion region analysis in the connected region analysis refers to removing pixels on the boundary, so as to eliminate noise or separate irrelevant small regions; and the open-close operation region analysis in the connected region analysis is a combination of dilation and erosion, the open operation can remove small isolated noise, and the close operation can fill small holes in the region.

[0064] The boundary curvature and concave-convex index are calculated for each spatial aggregation result, that is, the boundary shape of the abnormal region is analyzed, the boundary curvature measures the degree of boundary bending, and the concave-convex index indicates whether the boundary is convex or concave, and a boundary continuity analysis result is generated to evaluate the integrity and regularity of the boundary of the abnormal region.

[0065] The geometric morphological parameters, spectral feature parameters, and texture abnormality parameters are extracted using the spatial aggregation result and the boundary continuity analysis result, that is, the overall shape information, boundary characteristics, color spectrum information, and texture features of the abnormal region are combined to extract a parameter set that can accurately describe the defect, and provide a basis for subsequent defect classification and judgment.

[0066] Further, the application further comprises: obtaining a parameter mapping of the coil manufacturing parameters and the defect set; after obtaining the defect mask, performing associated matching of the manufacturing parameters according to the defect mask and the parameter mapping to generate an associated record; generating production feedback according to the associated record, and using the production feedback for coil production management.

[0067] Specifically, a parameter mapping of coil manufacturing parameters and defect set is obtained, which establishes a corresponding relationship between various process parameters, equipment settings and operating conditions involved in the production process of the coil and the detected defect types and their characteristics. Coil manufacturing parameters can include coil thickness, winding tension, current, heating temperature, etc., while defect set parameters include crack length, area, spectral difference and texture abnormality strength, etc. The parameter mapping can help analyze the production conditions of the coil that may cause a specific type of defect to occur.

[0068] After obtaining the defect mask, the associated matching of manufacturing parameters is performed according to the defect mask and the parameter mapping, and the associated record is generated, i.e. using the abnormal area marked in the defect mask, the abnormal characteristics are mapped to the production parameters, the specific process conditions or equipment states that may cause these defects are found out, and the structured record is formed.

[0069] Production feedback is generated according to the associated record, and coil production management is performed using the production feedback, i.e. the associated record is converted into an operable production improvement suggestion, such as adjusting tension, controlling temperature or optimizing winding speed, so as to reduce the occurrence of defects in subsequent production and improve product quality. Production management can include adjustment prompts of an automatic control system or operation suggestions for manual intervention.

[0070] In summary, the defect identification method of the coil automatic detection machine provided in the present application has the following technical effects: by achieving the technical target of high-precision, full-spectrum, multi-scale automatic identification and positioning of the surface defects of the coil, the technical effects of improving the accuracy of defect detection, enhancing the quantitative analysis ability of defects, realizing the closed-loop feedback of production parameters and optimizing the production management of the coil are achieved.

[0071] Embodiment two, based on the same inventive concept as the defect identification method of the coil automatic detection machine in the foregoing embodiments, the present application also provides a defect identification system of a coil automatic detection machine, please refer to the attached Figure 2, comprising: a detection instruction activation module 1 for triggering the detection instruction of the automatic detection machine of the coil when the coil reaches the target detection area; an image acquisition module 2 for controlling the visible light camera and the infrared camera arranged directly above the target detection area to perform image acquisition and establish a detection image set after starting the annular adjustable light source according to the detection instruction; an image traversal scanning module 3 for performing distortion and illumination correction on the detection image set, performing multi-spectral feature fusion on the corrected image set, constructing an enhanced image, inputting the enhanced image into a window recognition channel, performing image traversal scanning of the enhanced image by using a sliding window, calculating the texture energy and the direction consistency index in each sliding window, and establishing a coordinate index; an image anomaly comparison module 4 for calling a calibration position image according to the coordinate index, performing multi-scale image anomaly comparison on the enhanced image corresponding to the calibration position image and the coordinate index, and establishing a pixel point with an abnormal value identifier; and a defect mask output module 5 for performing morphological aggregation and boundary continuity analysis on the pixel point with the abnormal value identifier, extracting geometric morphological parameters, spectral feature parameters and texture anomaly parameters, and outputting a defect mask.

[0072] Further, the image traversal scanning module 3 is configured to perform the following operation steps: calling a three-dimensional calibration model of the target detection area from a calibration database of the detection machine, performing joint correction of radial distortion and tangential distortion on the visible light image and the infrared image respectively by using the three-dimensional calibration model; obtaining environmental spectrum monitoring data of the target detection area, constructing an illumination distribution surface model based on the environmental spectrum monitoring data, and performing illumination correction on the visible light image and the infrared image respectively after distortion correction by using the illumination distribution surface model under block brightness equalization; and performing the following: dividing the visible light image or the infrared image into N grid blocks, calculating a stable brightness statistic for each grid block; calculating the gain map of each pixel after configuring the target brightness constant according to the stable brightness statistic, performing multiplicative correction of the original image to complete the illumination correction.

[0073] Further, the image traversal scanning module 3 is configured to perform the following operation steps: performing spatial geometric registration of the visible light image and the infrared image after illumination correction at the pixel level, performing hierarchical feature decomposition, extracting low-frequency layer features, medium-frequency layer features and high-frequency layer features, and performing adaptive weight fusion based on the definition scores of the multi-layer features and the defect prior probability; and constructing an enhanced image by using the adaptive weight fusion result.

[0074] Further, the image traversal scanning module 3 is configured to perform the following steps: calling a texture extraction layer of the window recognition channel, performing an enhanced image intra-texture complexity evaluation, and establishing a texture complexity evaluation identifier; establishing a window shape library of a sliding window, the window shape library including a square window and a rectangular window, generating a position window shape constraint after shape matching of the window shape library using the texture complexity evaluation identifier; performing a texture energy descriptor calculation after configuring a multi-scale position window using the position window shape constraint, establishing a texture energy distribution, and establishing a coordinate index according to the texture energy distribution and a direction consistency index.

[0075] Further, the image traversal scanning module 3 is configured to perform the following steps: performing gradient direction clustering on each multi-scale position window, and calculating a main direction concentration degree; taking the main direction concentration degree as a direction consistency index, and establishing a coordinate index according to the texture energy distribution and the direction consistency index.

[0076] Further, the image anomaly comparison module 4 is configured to perform the following steps: calling a multi-scale calibration position image in a calibration library according to the coordinate index; performing set correction and spectral normalization processing on the multi-scale calibration position image, performing image registration at different scale layers, and performing anomaly comparison based on spectral difference, texture difference, and set difference to establish an anomaly value identifier.

[0077] Further, the defect mask output module 5 is configured to perform the following steps: performing spatial aggregation of abnormal pixels using a connected region analysis based on morphology, and merging adjacent similar abnormal pixel points, the connected region analysis including expansion region analysis, corrosion region analysis, and open-close operation region analysis; performing boundary curvature and concave-convex index calculation on each spatial aggregation result to generate a boundary continuity analysis result; and performing geometric morphological parameter, spectral feature parameter, and texture anomaly parameter extraction using the spatial aggregation result and the boundary continuity analysis result.

[0078] Further, the defect mask output module 5 is configured to perform the following steps: obtaining a parameter mapping of coil manufacturing parameters and a defect set; performing associated matching of manufacturing parameters according to the defect mask and the parameter mapping after obtaining the defect mask to generate an associated record; generating production feedback according to the associated record, and using the production feedback for coil production management.

[0079] Through the foregoing detailed description of the defect identification method of the coil automatic detection machine, those skilled in the art can clearly understand the defect identification system of the coil automatic detection machine in the embodiment, which is relatively simple because it corresponds to the method disclosed in the embodiment, and the relevant parts can be referred to the method part description.

[0080] In the third embodiment, based on the inventive concept of the defect identification method of the coil automatic detection machine in the foregoing embodiments, the application further provides an electronic device, comprising: at least one processor; a memory in communication connection with the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the steps of the defect identification method of the coil automatic detection machine in any one of the first embodiment.

[0081] The accompanying drawings are included to provide a further understanding of the application, and are incorporated herein and constitute a part of the detailed description. It will be apparent to those skilled in the art that the following examples can be practiced with the details as specific embodiments of the application. Figure 3 The accompanying drawings are included to provide a further understanding of the application, and are incorporated herein and constitute a part of the detailed description. It will be apparent to those skilled in the art that the following examples can be practiced with the details as specific embodiments of the application. Figure 3 In the third embodiment, based on the inventive concept of the defect identification method of the coil automatic detection machine in the foregoing embodiments, the application further provides an electronic device, comprising: at least one processor; a memory in communication connection with the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the steps of the defect identification method of the coil automatic detection machine in any one of the first embodiment.

[0082] The above description of disclosed embodiments provides enabling disclosure sufficient for one of ordinary skill in the art to practice or use the application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and generic principles defined herein can be applied to other embodiments without departing from the spirit or scope of the application. Thus, the present application is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

[0083] Obviously, various modifications and changes can be made to the present application by those skilled in the art without departing from the spirit and scope of the present application. Thus, it is intended that the present application cover the modifications and changes as long as they come within the scope of the application and its equivalents.

Claims

1. A defect recognition method of a coil automatic inspection machine, characterized by, The method comprises: When the coil reaches the target detection area, a detection instruction of the coil automatic detection machine is triggered; After starting the annular adjustable light source according to the detection instruction, a visible light camera and an infrared camera arranged directly above the target detection area are controlled to perform image acquisition, and a detection image set is established; After distortion and illumination correction are performed on the detection image set, multi-spectrum feature fusion is performed on the corrected image set to construct an enhanced image, and the enhanced image is input into a window recognition channel, image traversal scanning of the enhanced image is performed by using a sliding window, texture energy and direction consistency indexes in each sliding window are calculated, and a coordinate index is established; According to the coordinate index, a calibration position image is called, multi-scale image anomaly comparison is performed on the calibration position image and the enhanced image corresponding to the coordinate index, and a pixel point with an anomaly value identifier is established; After morphological aggregation and boundary continuity analysis are performed on the pixel point with the anomaly value identifier, geometric morphological parameters, spectral feature parameters and texture anomaly parameters are extracted, and a defect mask is output.

2. The defect recognition method of a coil automatic inspection machine according to claim 1, wherein The distortion and illumination correction performed on the detection image set comprises: A three-dimensional calibration model of the target detection area is called from a calibration database of the detection machine, radial distortion and tangential distortion joint correction are performed on the visible light image and the infrared image respectively by using the three-dimensional calibration model; Environmental spectrum monitoring data of the target detection area are acquired, a light distribution surface model is constructed based on the environmental spectrum monitoring data, and illumination correction is performed on the visible light image and the infrared image respectively after distortion correction by using the light distribution surface model under block brightness equalization, and the following is performed: The visible light image or the infrared image is divided into N grid blocks, and a stable brightness statistic is calculated for each grid block; After a target brightness constant is configured according to the stable brightness statistic, a gain map of each pixel is calculated, multiplicative correction of the original image is performed, and the illumination correction is completed.

3. The defect recognition method of a coil automatic inspection machine according to claim 2, wherein The multi-spectrum feature fusion performed on the corrected image set to construct the enhanced image comprises: After spatial geometric registration is performed on the visible light image and the infrared image after illumination correction at a pixel level, hierarchical feature decomposition is performed, low-frequency layer features, medium-frequency layer features and high-frequency layer features are extracted, and adaptive weight fusion is performed based on the definition scores of the multi-layer features and defect prior probabilities; The adaptive weight fusion result is used to construct the enhanced image.

4. The defect recognition method of a coil automatic inspection machine according to claim 1, wherein The input of the enhanced image into the window recognition channel and the image traversal scanning of the enhanced image by using the sliding window comprise: The texture extraction layer of the window recognition channel is called, texture complexity evaluation in the enhanced image is performed, and a texture complexity evaluation identifier is established; A window shape library of the sliding window is established, the window shape library comprises a square window and a rectangular window, a position window shape constraint is generated after shape matching of the window shape library is performed by using the texture complexity evaluation identifier; After a multi-scale position window is configured by using the position window shape constraint, texture energy descriptor calculation is performed, a texture energy distribution is established, and a coordinate index is established according to the texture energy distribution and the direction consistency index.

5. The defect recognition method of a coil automatic inspection machine according to claim 4, wherein The establishment of the coordinate index according to the texture energy distribution and the direction consistency index comprises: perform gradient direction clustering on each multi-scale position window, and calculate a main direction concentration degree; use the main direction concentration degree as a direction consistency index, and establish a coordinate index according to the texture energy distribution and the direction consistency index.

6. The defect recognition method of a coil automatic inspection machine according to claim 1, wherein The multi-scale image anomaly comparison of the enhanced image corresponding to the calibration position image and the coordinate index includes: calling a multi-scale calibration position image in a calibration library according to the coordinate index; performing set correction and spectral normalization processing on the multi-scale calibration position image, performing image registration at different scale layers, and performing anomaly comparison based on spectral difference, texture difference and set difference to establish an abnormal value identifier.

7. The defect recognition method of a coil automatic inspection machine according to claim 1, wherein The morphological aggregation and boundary continuity analysis of the pixel points with the abnormal value identifier include: performing spatial aggregation of abnormal pixels by using a morphological connected region analysis, and merging adjacent similar abnormal pixel points, the connected region analysis including dilation region analysis, erosion region analysis and open-close operation region analysis; calculating a boundary curvature and concave-convex index for each spatial aggregation result to generate a boundary continuity analysis result; extracting geometric morphological parameters, spectral feature parameters and texture anomaly parameters by using the spatial aggregation result and the boundary continuity analysis result.

8. The defect recognition method of a coil automatic inspection machine according to claim 1, wherein After the defect mask is output, the following includes: obtaining a parameter mapping of a coil manufacturing parameter and a defect set; after obtaining the defect mask, performing associated matching of the manufacturing parameter according to the defect mask and the parameter mapping to generate an associated record; generating production feedback according to the associated record, and using the production feedback for coil production management.

9. A defect recognition system of a coil automatic inspection machine, characterized by, A defect identification method of a coil automatic detection machine for implementing any one of claims 1-8, the system comprising: a detection instruction activation module configured to trigger a detection instruction of the coil automatic detection machine when a coil reaches a target detection area; an image acquisition module configured to control a visible light camera and an infrared camera arranged directly above the target detection area to perform image acquisition and establish a detection image set after starting a ring-shaped adjustable light source according to the detection instruction; an image traversal scanning module configured to perform distortion and illumination correction on the detection image set, perform multi-spectral feature fusion on the corrected image set to construct an enhanced image, input the enhanced image into a window recognition channel, perform image traversal scanning of the enhanced image by using a sliding window, calculate texture energy and direction consistency indexes in each sliding window, and establish a coordinate index; an image anomaly comparison module configured to call a calibration position image according to the coordinate index, perform multi-scale image anomaly comparison of an enhanced image corresponding to the calibration position image and the coordinate index, and establish pixel points with an abnormal value identifier; a defect mask output module configured to perform morphological aggregation and boundary continuity analysis on the pixel points with the abnormal value identifier, extract geometric morphological parameters, spectral feature parameters and texture anomaly parameters, and output a defect mask.

10. An electronic device, comprising: comprise: at least one processor; a memory in communication with the at least one processor; The memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the steps of the defect identification method of the coil automatic detection machine according to any one of claims 1 to 8.

Images