A method and system for detecting defects in a glass fiber mesh

Abstract

The application discloses a kind of glass fibre mesh cloth defect detection method and system, it is related to intelligent detection related technical field, this method includes: by setting shooting parameter is moved to the target glass fibre mesh cloth and is photographed, obtains detection target image information.Acquisition is placed background image, and the coupling feature analysis is carried out to placed background image, determines coupling influence parameter.Based on coupling influence parameter, background decoupling is carried out to detection target image information, and target image is extracted.Target image is meshed and segmented, and image mesh segmentation library is obtained.Extract defect feature is used to traverse image mesh segmentation library, and defect detection result is obtained.The technical problem that the glass fibre mesh cloth defect detection method in prior art is difficult to execute detection task under complex environment, leading to insufficient detection accuracy and reliability is solved.High-efficiency, accurate detection glass fibre mesh cloth defect is realized, and the technical effect that the accuracy of detection and the adaptability of system are greatly improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of intelligent detection technology, and particularly relates to a defect detection method and system for glass fiber mesh cloth. BACKGROUND

[0002] With the continuous development of industrial production, glass fiber mesh cloth is widely used in many fields such as construction, automobile and electronics, and its quality directly affects the performance and service life of the product. However, glass fiber mesh cloth is prone to various defects such as cracks, bubbles and color difference during production, which not only affects the appearance but also may reduce the mechanical strength and durability of the material. Traditional defect detection methods mainly rely on manual visual inspection or simple image processing technology, which has the problems of low detection efficiency, poor accuracy, easy to be disturbed by background light and mesh structure, etc.

[0003] Therefore, in the prior art, the defect detection method for glass fiber mesh cloth cannot perform detection tasks in complex environments, resulting in insufficient detection accuracy and reliability. SUMMARY

[0004] The present application provides a defect detection method and system for glass fiber mesh cloth, which solves the technical problem that the defect detection method for glass fiber mesh cloth in the prior art cannot perform detection tasks in complex environments, resulting in insufficient detection accuracy and reliability. Through multi-scale mesh segmentation and coupling influence parameter background decoupling, efficient and accurate detection of glass fiber mesh cloth defects is achieved, greatly improving the detection accuracy and adaptability of the system.

[0005] The present application provides a defect detection method for glass fiber mesh cloth, which comprises: moving the target glass fiber mesh cloth by setting the shooting parameters to obtain the detection target image information, the shooting parameters including shooting distance, shooting angle and shooting parameters, wherein the shooting parameters meet the preset shooting requirements; obtaining a placement background image, performing coupling feature analysis on the placement background image to determine the coupling influence parameters; based on the coupling influence parameters, decoupling the background of the detection target image information to extract the target image; performing mesh segmentation on the target image to obtain an image mesh segmentation library; using the pre-extracted defect features to traverse the image mesh segmentation library to obtain the defect detection result.

[0006] In the implementation method, before the target glass fiber mesh cloth is moved and photographed by setting shooting parameters, it includes: analyzing the shooting light source characteristics, detection target motion characteristics, and shooting equipment setting constraint characteristics of the detection environment to obtain environmental acquisition characteristics; extracting shooting influencing parameters based on the environmental acquisition characteristics, performing compensation analysis based on the shooting influencing parameters, and obtaining compensation target parameters; using the compensation target parameters as the setting target, performing relevant compensation calculations on the shooting distance, shooting angle, and shooting parameters of the shooting equipment, and obtaining the shooting parameters with the preset shooting requirements as constraints.

[0007] In an implementation, after acquiring the shooting parameters, the method further includes: setting a device movement speed according to the target motion characteristics; and configuring a shooting frame rate according to the device movement speed.

[0008] In the implementation method, the coupling feature analysis is performed on the placed background image to determine the coupling influence parameters, including: extracting the texture features, color distribution features, and illumination distribution features of the placed background image to obtain the texture features, color distribution features, and illumination distribution features of the background image; extracting the texture features, color distribution features, and illumination distribution features of the target glass fiber mesh cloth to obtain the texture features, color distribution features, and illumination distribution features of the background image of the target mesh cloth; analyzing the differences and associated influences of the image feature parameters of the background image features and the target mesh cloth based on the texture features, color distribution features, and illumination distribution features; obtaining background decoupling image features based on the differences in the image feature parameters; and obtaining the coupling influence parameters based on the associated influences of the texture features, color distribution features, and illumination distribution features.

[0009] In the implementation method, the background of the detected target image information is decoupled based on the coupling influence parameters to extract the target image, including: extracting key frames from the detected target image information, annotating the key frames according to the positioning coordinates, splicing the glass fiber mesh cloth unit image based on the positioning coordinates, the glass fiber mesh cloth unit image being the image information corresponding to the unit identification structure of the glass fiber mesh cloth with an edge boundary; removing background features of the glass fiber mesh cloth unit image according to the background decoupling image features to obtain initial target image information; constructing a coupling matrix based on the coupling influence parameters; and using the coupling matrix to decouple the background parameters of the initial target image information, restoring the image feature parameters of the target mesh cloth, and obtaining the target image.

[0010] In the implementation method, the target image is grid-segmented to obtain an image grid segmentation library, including: obtaining the proportional relationship between the defect distribution size and the mesh size according to the defect type combined with the pre-extracted defect feature library, and constructing a scale segmentation list of multiple defect types and defect levels; obtaining a detection accuracy scale constraint; matching and extracting the scale segmentation list according to the detection accuracy scale constraint to obtain a multi-scale segmentation target; grid-segmenting the target image according to the multi-scale segmentation target to obtain multiple image grid segmentation libraries, each image grid segmentation library having a scale identifier; searching and matching the scale identifier with the scale segmentation list to obtain the corresponding defect type and defect level to identify the defect type of each image grid segmentation library.

[0011] In the implementation method, the pre-extracted defect features are used to traverse the image grid segmentation library to obtain the defect detection results, including: using the defect type identification of the image grid segmentation library and the pre-extracted defect features to match and extract to obtain the target defect features; using the target defect features to traverse each segmentation grid feature of the image grid segmentation library, and when the match with the target defect features is successful, the defect detection results are obtained.

[0012] In the implementation method, the target image is grid-segmented according to the multi-scale segmentation target, including: grid-segmenting the target image according to the multi-scale segmentation target to obtain segmentation blocks and segmentation edges, wherein the segmentation blocks are stored as image grids; boundary neighborhood expansion of the segmentation edges is performed to obtain an extended grid; regional features of the extended grid are compared with boundary region features of the segmentation units of the segmentation edges, and when the boundary region features have a continuation relationship in the extended grid, it is determined that a cross-border defect exists; with the segmentation edge as the center, edge compensation segmentation is performed according to the corresponding scale segmentation target to obtain a compensated segmentation block and add it to the image grid segmentation library.

[0013] The present application also provides a defect detection system for glass fiber mesh cloth, including: an image acquisition module, used to perform mobile shooting of the target glass fiber mesh cloth by setting shooting parameters to obtain detection target image information, the shooting parameters including shooting distance, shooting angle, and shooting parameters, wherein the shooting parameters meet preset shooting requirements; a coupling influence acquisition module, used to obtain a placement background image, perform coupling feature analysis on the placement background image, and determine coupling influence parameters; a background decoupling module, used to perform background decoupling on the detection target image information based on the coupling influence parameters to extract the target image; a grid segmentation module, used to perform grid segmentation on the target image to obtain an image grid segmentation library; a detection result acquisition module, used to traverse the image grid segmentation library using pre-extracted defect features to obtain defect detection results.

[0014] The present application proposes a method and system for defect detection of glass fiber mesh cloth. By setting shooting parameters, the target glass fiber mesh cloth is moved and photographed to obtain detection target image information. The shooting parameters include shooting distance, shooting angle, and shooting parameters, wherein the shooting parameters meet preset shooting requirements; a placement background image is obtained, coupling feature analysis is performed on the placement background image, and coupling influence parameters are determined; background decoupling of the detection target image information is performed based on the coupling influence parameters to extract the target image; grid segmentation is performed on the target image to obtain an image grid segmentation library; and the image grid segmentation library is traversed using pre-extracted defect features to obtain defect detection results. This solves the technical problem that the existing glass fiber mesh cloth defect detection method is difficult to perform detection tasks in complex environments, resulting in insufficient detection accuracy and reliability. Through multi-scale grid segmentation and background decoupling of coupling influence parameters, efficient and accurate detection of glass fiber mesh cloth defects is achieved, which greatly improves the technical effect of detection accuracy and system adaptability. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] To more clearly illustrate the technical solutions of the embodiments of the present disclosure, the accompanying drawings of the embodiments of the present disclosure are briefly introduced below. Flowcharts are used in this application to illustrate the operations performed by the systems according to the embodiments of the present application. It should be understood that the preceding or following operations are not necessarily performed in precise order. Instead, various steps may be processed in reverse order or simultaneously as needed. Furthermore, other operations may be added to these processes, or one or more operations may be removed from these processes.

[0016] Figure 1 A schematic flow chart of a defect detection method for a glass fiber mesh cloth provided in an embodiment of the present application.

[0017] Figure 2 A schematic structural diagram of a glass fiber mesh defect detection system provided in an embodiment of the present application.

[0018] Explanation of reference numerals: image acquisition module 11 , coupling influence acquisition module 12 , background decoupling module 13 , grid segmentation module 14 , detection result acquisition module 15 . DETAILED DESCRIPTION

[0019] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below.

[0020] In order to make the purpose, technical solutions and advantages of this application clearer, the application will be further described in detail below with reference to the accompanying drawings. The described embodiments should not be regarded as limiting this application. All other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application.

[0021] In the following description, reference is made to “some embodiments”, which describes a subset of all possible embodiments, but it will be understood that “some embodiments” may be the same subset or different subsets of all possible embodiments and may be combined with each other without conflict, and the terms “first\second” involved are merely used to distinguish similar objects and do not represent a specific ordering of the objects. The terms “including” and “having” and any variations are intended to cover non-exclusive inclusions. For example, a process, method, system, product, or server that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or modules that are not clearly listed or that are inherent to these processes, methods, products, or devices. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art to which this application belongs. The terms used herein are for the purpose of describing the embodiments of this application only.

[0022] The present invention provides a method and system for detecting defects in a glass fiber mesh cloth. Figure 1 As shown, the method includes:

[0023] The target fiberglass mesh is photographed by moving the target with set shooting parameters, including shooting distance, shooting angle, and shooting parameters, where the shooting parameters meet preset shooting requirements. A background image is obtained, and coupling characteristics are analyzed on the background image to determine coupling influencing parameters.

[0024] With the continuous development of industrial production, glass fiber mesh is widely used in construction, automobile, electronics and other fields, its quality directly affects the performance and service life of the product. However, glass fiber mesh is prone to produce various defects in the production process, such as cracks, bubbles, color difference, etc. These defects not only affect the appearance, but also may reduce the mechanical strength and durability of the material. The traditional defect detection method mainly relies on manual visual inspection or simple image processing technology, these methods have low detection efficiency, poor accuracy, are easy to be disturbed by background light and mesh structure, etc. The target glass fiber mesh is moved and photographed by setting the shooting parameters, and the target image information is obtained. The shooting parameters include shooting distance, shooting angle, shooting parameters, wherein the shooting parameters meet the preset shooting requirements, and the preset shooting requirements are pre-set shooting requirements such as shooting area requirement, image brightness requirement, etc. The placement background image is obtained, which refers to the background image in the shooting environment without the target glass fiber mesh, so as to facilitate the subsequent analysis of the interference of the background on the detection result. The coupling feature analysis is performed on the placement background image, and the coupling influence parameter is determined, which is the feature with strong correlation between the placement background image and the target image. Since the correlation between the two features is strong, it is difficult to remove, so the feature with strong correlation is extracted for subsequent targeted decoupling removal.

[0025] The method provided by the embodiment of the application further includes: performing shooting light source feature analysis, detection target motion feature analysis, and shooting device setting constraint feature analysis on the detection environment to obtain environment collection features. A shooting influence parameter is extracted according to the environment collection features, and compensation analysis is performed according to the shooting influence parameter to obtain a compensation target parameter. The shooting distance, shooting angle, and shooting parameter of the shooting device are compensated and calculated with the compensation target parameter as the setting target, and the shooting parameter is obtained with the preset shooting requirement as the constraint condition.

[0026] Before the target glass fiber mesh is moved and photographed by setting the shooting parameters, the detection environment needs to be analyzed before the shooting parameters are set and the moving photographing is performed, including the analysis of the shooting light source feature, the detection target motion feature, and the shooting device setting constraint feature to obtain the environment collection features. The shooting light source feature refers to the brightness data of the light source in the environment. The detection target motion feature refers to the moving speed, direction and other motion parameters of the glass fiber mesh in the detection process. The shooting device setting constraint feature refers to the working limitation of the shooting device in the corresponding environment, such as the physical space limitation and angle limitation of the device. The environment collection features are obtained according to the analysis results of the features.

[0027] Furthermore, shooting influencing parameters are extracted based on the environmental acquisition features. The shooting influencing parameters include: brightness parameter, distance parameter, setting angle parameter, area parameter, etc. Compensation analysis is performed based on the shooting influencing parameters to obtain compensation target parameters. The compensation target parameter is the difference between the standard shooting influencing parameter and the shooting influencing parameter. Taking the distance parameter as an example, the vertical distance between the camera and the detection target is 37cm, while the standard distance in the standard shooting influencing parameter is 45cm. In this case, the corresponding distance in the compensation target parameter is +8cm. Furthermore, using the compensation target parameter as the setting target, relevant compensation calculations are performed on the shooting distance, shooting angle, and shooting parameters of the shooting device. Since the shooting distance and shooting angle both contain corresponding compensation parameters in the compensation target parameters, the shooting device can be directly compensated. The core of the compensation calculation is the calculation of shooting parameters, such as the calculation of light intensity and aperture, shooting distance and focal length, etc. Taking light intensity as an example, if the ambient light intensity is to be increased or decreased, the camera aperture needs to be adjusted. For example, if the ambient light brightness is to be increased by 20%, the corresponding aperture area needs to be increased by 20%. Based on the inverse relationship between the aperture area and the square of the f-number, the corresponding aperture reduction ratio of approximately 9.54% is obtained. Combined with the camera's current aperture setting, the new aperture f-number is obtained. The preset shooting requirements are pre-set shooting requirements such as shooting area requirements and image brightness requirements. After completing the compensation calculation, it is determined whether the captured image obtained after compensating the shooting parameters falls within the preset shooting requirements. If so, the shooting parameters that meet the requirements are obtained.

[0028] The method provided in the embodiment of the present application further includes: setting a device movement speed according to the target motion characteristics, and configuring a shooting frame rate according to the device movement speed.

[0029] After obtaining the shooting parameters, the method further includes: obtaining the target's movement speed based on the target's motion characteristics, and obtaining the device's movement speed based on the target's movement speed combined with a shooting movement speed standard, wherein the shooting movement speed standard is a preset relative movement speed between the camera and the fiberglass mesh. The device's movement speed is the difference between the shooting movement speed standard and the target's movement speed. A shooting frame rate is configured based on the device's movement speed and the target's motion characteristics. The shooting frame rate is the optimal shooting frame rate for image resolution under the shooting movement speed standard.

[0030] The method provided in the embodiment of the present application also includes: extracting texture features, color distribution features, and illumination distribution features of the placed background image to obtain the texture features, color distribution features, and illumination distribution features of the background image. Extracting texture features, color distribution features, and illumination distribution features of the target glass fiber mesh cloth to obtain the texture features, color distribution features, and illumination distribution features of the background image of the target mesh cloth. Based on the texture features, color distribution features, and illumination distribution features, respectively, analyzing the differences and associated influences of the image feature parameters of the background image features and the target mesh cloth. Based on the differences in the image feature parameters, background decoupling image features are obtained. Based on the associated influences of the texture features, color distribution features, and illumination distribution features, the coupling influence parameters are obtained.

[0031] A coupling feature analysis is performed on the placed background image to determine coupling influence parameters, including the following steps: Before photographing the target glass fiber mesh cloth, a placed background image, i.e., a background image without the target mesh cloth, is first acquired. Texture features, color distribution features, and illumination distribution features are then extracted from the background image to facilitate subsequent analysis of the background's impact on the target image. Texture features describe the details and structure of the image surface, including patterns, roughness, and other information. Common texture feature extraction methods include gray-level co-occurrence matrices and local binary patterns. Color distribution features describe the distribution of colors in the image, including color histograms and features after color space conversion. Illumination distribution features describe the distribution of illumination in the image, including the spatial distribution of light intensity, the location of the light source, and its range of influence. Texture, color, and illumination distribution features are extracted from the target glass fiber mesh cloth in the same manner to obtain the texture, color, and illumination distribution features of the background image of the target mesh cloth. Furthermore, based on the texture features, color distribution features, and illumination distribution features, the differences and correlation influences between the background image features and the target mesh image are analyzed. Image feature parameter differences refer to features with significant differences, while correlation influences refer to features with high similarity. The differences and correlation influences can be determined by applying a threshold to the difference between the feature parameters of the background and target mesh images. Based on the image feature parameter differences, image features solely attributable to the background are extracted and used as the basis for background decoupling, thereby obtaining background-decoupled image features. The background-decoupled image features are then used to directly remove corresponding features from the target glass fiber mesh image to reduce background interference. Based on the correlation influences of the texture features, color distribution features, and illumination distribution features, corresponding features of the correlation influences are obtained. Since correlation influences are difficult to distinguish within an image, image settings can be adjusted to differentiate the correlation influences or compensate for the influencing features, thereby obtaining coupling influence parameters. Coupling influence parameters are specific parameters that describe the influence of background features (such as illumination and color) on target image features. These parameters typically include: Light intensity adjustment factor: describes the brightness or light intensity ratio that needs to be adjusted. Color deviation adjustment factor: describes the degree to which the color needs to be shifted, such as the value of the shift toward cooler or warmer colors. Contrast adjustment factor: describes the ratio by which the contrast needs to be increased or decreased.

[0032] The target image information is background-decoupled based on the coupling influence parameter to extract the target image. The target image is grid-segmented to obtain an image grid segmentation library. The pre-extracted defect features are used to traverse the image grid segmentation library to obtain a defect detection result.

[0033] Based on the coupling influence parameters, the detection target image information is subjected to background decoupling, so that the features with strong correlation in the background are decoupled in the detection target image information, thereby obtaining the decoupled target image. Further, the target image is grid-segmented to obtain an image grid segmentation library. The image grid segmentation library contains corresponding defect type identifiers at different scales. Finally, the image grid segmentation library is traversed using the pre-extracted defect features, that is, the segmentation results at the corresponding scale in the image grid segmentation library are identified according to the pre-extracted defect features and the corresponding defect types, thereby obtaining defect detection results. This solves the technical problem that the glass fiber mesh cloth defect detection method in the prior art is difficult to perform detection tasks in complex environments, resulting in insufficient detection accuracy and reliability. Through multi-scale grid segmentation and coupling influence parameter background decoupling, efficient and accurate detection of glass fiber mesh cloth defects is achieved, which greatly improves the technical effect of detection accuracy and system adaptability.

[0034] The method provided in an embodiment of the present application further includes: extracting key frames from the detected target image information, annotating the key frames according to the positioning coordinates, and splicing a unit image of the glass fiber mesh cloth based on the positioning coordinates, wherein the unit image of the glass fiber mesh cloth is image information corresponding to a unit identification structure of the glass fiber mesh cloth having an edge boundary. Background features of the unit image of the glass fiber mesh cloth are removed based on the background decoupling image features to obtain initial target image information. A coupling matrix is ​​constructed based on the coupling influence parameters. The coupling matrix is ​​used to decouple the background parameters of the initial target image information, and the image feature parameters of the target mesh cloth are restored to obtain the target image.

[0035] Background decoupling of the detected target image information is performed based on the coupling influence parameters to extract the target image. This includes: before completing background decoupling, key frames are first extracted from the continuous image sequence to obtain representative or significantly changed frames in the image sequence, thereby reducing the amount of data to be processed while retaining important information. Subsequently, the key frames are marked based on the positioning coordinates in the image, their positions within the overall image are determined, and the key frames are annotated. The positioning coordinates are specific positioning points in the image, and the positioning coordinates of the corresponding image are determined based on these positioning coordinates. Then, based on these positioning coordinates, the unit images of the fiberglass mesh cloth are spliced ​​together to form a complete mesh cloth image. The fiberglass mesh cloth unit image is image information corresponding to the unit identification structure of the fiberglass mesh cloth with edge boundaries. Using image processing technology, the background decoupling image features in the fiberglass mesh cloth unit image are directly removed to eliminate background interference with the target image, thereby obtaining initial target image information. Furthermore, a coupling matrix is ​​constructed using the determined coupling influence parameters. The coupling matrix is ​​used to group multiple coupling-influencing parameters together so that these adjustments can be uniformly applied within the image processing algorithm. It describes the interplay between background and target image features and is a key parameter in the background decoupling process. By applying the coupling matrix to the initial target image information, the background parameters are decoupled, the image characteristic parameters of the target mesh are restored, and the true image characteristics of the target mesh are recovered to obtain the target image.

[0036] The method provided in the embodiment of the present application also includes: obtaining the proportional relationship between the defect distribution size and the mesh size based on the defect type combined with the pre-extracted defect feature library, and constructing a scale segmentation list of multiple defect types and defect levels. Obtaining a detection accuracy scale constraint. Matching and extracting the scale segmentation list based on the detection accuracy scale constraint to obtain a multi-scale segmentation target. Grid-segmenting the target image according to the multi-scale segmentation target to obtain multiple image grid segmentation libraries, each of which has a scale identifier. Searching and matching the scale identifier with the scale segmentation list to obtain the corresponding defect type and defect level to identify the defect type of each image grid segmentation library.

[0037] Based on a pre-extracted defect feature library containing the corresponding features of each defect, the proportional relationship between the distribution size of different defect types and mesh size is analyzed to construct a scale segmentation list for multiple defect types and grades. This list is used to guide the subsequent image segmentation process, ensuring that defects of different types and grades can be accurately identified at the appropriate scale. Subsequently, based on the requirements of the inspection system and the application scenario, scale constraints for detection accuracy are set to obtain the detection accuracy scale constraints. Matching and extraction are performed on the scale segmentation list according to the detection accuracy scale constraints, screening out multi-scale segmentation targets that meet the inspection requirements. Multi-scale segmentation targets refer to image segmentation operations performed within different scale ranges to accommodate the detection needs of different defect types and grades. Based on the selected multi-scale segmentation targets, the target image is grid-segmented to generate multiple image grid segmentation libraries with scale identifiers. Each segmentation library corresponds to a specific scale range, facilitating subsequent defect type identification and detection. Finally, a search and match is performed between the scale identifiers and the scale segmentation list to obtain the corresponding defect type and grade, and then the defect type is identified for each image grid segmentation library. Complete accurate defect type identification for grid cells within different scale ranges, providing a basis for subsequent inspection and recording.

[0038] The method provided in the embodiment of the present application further includes: using the defect type identifier of the image grid segmentation library to match and extract the pre-extracted defect features to obtain target defect features. Using the target defect features, traversing each segmentation grid feature of the image grid segmentation library, and obtaining the defect detection result when a match with the target defect feature is successful.

[0039] The method uses the pre-extracted defect features to traverse the image grid segmentation library to obtain a defect detection result, including: using the defect type identifier in the image grid segmentation library to match and extract the pre-extracted defect features, obtaining the defect features corresponding to the defect type identifier in the image grid segmentation library, and obtaining the target defect features. Subsequently, using the target defect features to traverse each segmented grid feature in the image grid segmentation library, and when a match with the target defect feature is successful, the defect detection result is obtained.

[0040] The method provided in the embodiment of the present application also includes: performing grid segmentation on the target image according to the multi-scale segmentation target to obtain segmentation blocks and segmentation edges, wherein the segmentation blocks are stored as image grids. The boundary neighborhood of the segmentation edge is expanded to obtain an extended grid. The regional features of the extended grid are compared with the boundary region features of the segmentation unit of the segmentation edge, and when the boundary region features have a continuation relationship in the extended grid, it is determined that there is a cross-border defect. With the segmentation edge as the center, edge compensation segmentation is performed according to the corresponding scale segmentation target to obtain a compensated segmentation block and add it to the image grid segmentation library.

[0041] Grid segmentation of the target image is performed according to the multi-scale segmentation target, including: grid segmentation of the target image according to the multi-scale segmentation target to obtain segmentation blocks and segmentation edges, wherein the segmentation blocks are stored as image grids. The multi-scale segmentation target is a plurality of scale ranges set according to different defect types and levels, which are used to guide the grid segmentation of the image at different scales. The segmentation edges generated during the segmentation process are extended by boundary neighborhood to form an extended grid. The new grid area formed by boundary neighborhood expansion includes pixels of the original segmentation edge and its adjacent area. The extended grid is used to detect cross-border defects, i.e., defects that may span multiple grid blocks. These cross-border features are captured by extending the neighborhood. The regional features in the extended grid are compared with the boundary region features of the original segmentation edge. If the boundary region features in the extended grid are found to be spatially continuous and consistent, it indicates that the defect spans multiple grid blocks. A cross-border defect is then determined to be present. Cross-border defects refer to defects that span multiple grid blocks and require special processing to avoid missed detection. Finally, for segmented edges identified as cross-border defects, edge compensation segmentation is performed based on the scale segmentation target. This involves re-meshing the center of the segmented edges to generate compensated segments. These compensated segments more accurately cover the cross-border defect area, thereby improving the comprehensiveness and accuracy of defect detection. These compensated segments are then added to the corresponding image mesh segmentation library for subsequent defect detection.

[0042] In the above, refer to Figure 1 A method for detecting defects in a glass fiber mesh cloth according to an embodiment of the present invention is described in detail. Figure 2 A defect detection system for glass fiber mesh cloth according to an embodiment of the present invention is described.

[0043] A glass fiber mesh defect detection system according to an embodiment of the present invention solves the technical problem that existing glass fiber mesh defect detection methods have difficulty performing detection tasks in complex environments, resulting in insufficient detection accuracy and reliability. Through multi-scale grid segmentation and background decoupling of coupling influence parameters, efficient and accurate detection of glass fiber mesh defects is achieved, significantly improving the accuracy of detection and the adaptability of the system. A glass fiber mesh defect detection system includes: an image acquisition module 11, a coupling influence acquisition module 12, a background decoupling module 13, a grid segmentation module 14, and a detection result acquisition module 15.

[0044] The image acquisition module 11 is used to perform mobile shooting of the target glass fiber mesh cloth by setting shooting parameters to obtain detection target image information. The shooting parameters include shooting distance, shooting angle, and shooting parameters, wherein the shooting parameters meet preset shooting requirements.

[0045] The coupling influence acquisition module 12 is used to acquire a placement background image, perform coupling feature analysis on the placement background image, and determine coupling influence parameters.

[0046] The background decoupling module 13 is configured to perform background decoupling on the detection target image information based on the coupling influence parameter to extract the target image.

[0047] The grid segmentation module 14 is used to perform grid segmentation on the target image to obtain an image grid segmentation library.

[0048] The detection result acquisition module 15 is used to traverse the image grid segmentation library using the pre-extracted defect features to obtain the defect detection results.

[0049] Below, the specific configuration of the image acquisition module 11 will be described in detail. The image acquisition module 11 may further include: before the target glass fiber mesh cloth is moved and photographed by setting shooting parameters, it includes: performing shooting light source characteristics, detection target motion characteristics, and shooting equipment setting constraint characteristics analysis on the detection environment to obtain environmental acquisition characteristics. Shooting influencing parameters are extracted based on the environmental acquisition characteristics, and compensation analysis is performed based on the shooting influencing parameters to obtain compensation target parameters. With the compensation target parameters as the setting target, relevant compensation calculations are performed on the shooting distance, shooting angle, and shooting parameters of the shooting equipment, and the shooting parameters are obtained with the preset shooting requirements as constraints. Below, the specific configuration of the image acquisition module 11 will continue to be described in detail. The image acquisition module 11 further includes: setting the device movement speed based on the target motion characteristics. Configuring the shooting frame rate based on the device movement speed.

[0050] The specific configuration of the coupling influence acquisition module 12 will be described in detail below. The coupling influence acquisition module 12 may further include: performing coupling feature analysis on the placed background image to determine coupling influence parameters, including: extracting texture features, color distribution features, and illumination distribution features from the placed background image to obtain the texture features, color distribution features, and illumination distribution features of the background image. Extracting texture features, color distribution features, and illumination distribution features from the target glass fiber mesh cloth to obtain the texture features, color distribution features, and illumination distribution features of the background image of the target mesh cloth. Based on the texture features, color distribution features, and illumination distribution features, the differences and associated influences of the image feature parameters of the background image and the target mesh cloth are analyzed respectively. Based on the differences in the image feature parameters, background decoupling image features are obtained. Based on the associated influences of the texture features, color distribution features, and illumination distribution features, the coupling influence parameters are obtained.

[0051] Below, the specific configuration of the background decoupling module 13 will be described in detail. The background decoupling module 13 further includes: performing background decoupling on the detection target image information based on the coupling influence parameters, and extracting the target image, including: extracting key frames from the detection target image information, and marking the key frames according to the positioning coordinates, splicing the glass fiber mesh cloth unit image based on the positioning coordinates, and the glass fiber mesh cloth unit image is the image information corresponding to the unit identification structure of the glass fiber mesh cloth with edge boundaries. According to the background decoupling image features, the background features of the glass fiber mesh cloth unit image are eliminated to obtain the initial target image information. A coupling matrix is ​​constructed based on the coupling influence parameters. The coupling matrix is ​​used to decouple the background parameters of the initial target image information, restore the image feature parameters of the target mesh cloth, and obtain the target image.

[0052] The specific configuration of the grid segmentation module 14 will be described in detail below. The grid segmentation module 14 further includes: performing grid segmentation on the target image to obtain an image grid segmentation library, including: obtaining the proportional relationship between the defect distribution size and the mesh size according to the defect type combined with the pre-extracted defect feature library, and constructing a scale segmentation list of multiple defect types-defect levels. Obtain a detection accuracy scale constraint. Match and extract the scale segmentation list according to the detection accuracy scale constraint to obtain a multi-scale segmentation target. Grid segment the target image according to the multi-scale segmentation target to obtain multiple image grid segmentation libraries, each of which has a scale identifier. Search and match the scale identifier with the scale segmentation list to obtain the corresponding defect type and defect level to identify the defect type of each image grid segmentation library.

[0053] The specific configuration of the detection result acquisition module 15 will be described in detail below. The detection result acquisition module 15 further includes: utilizing pre-extracted defect features to traverse the image grid segmentation library to obtain defect detection results, including: utilizing the defect type identifiers in the image grid segmentation library to match and extract the pre-extracted defect features to obtain target defect features. The target defect features are then used to traverse each segmented grid feature in the image grid segmentation library. Upon successful matching with the target defect features, the defect detection result is obtained.

[0054] The specific configuration of the grid segmentation module 14 will be described in detail below. The grid segmentation module 14 further includes: grid segmenting the target image according to the multi-scale segmentation target, including: grid segmenting the target image according to the multi-scale segmentation target, obtaining segmentation blocks and segmentation edges, wherein the segmentation blocks are stored as image grids. The boundary neighborhood of the segmentation edge is expanded to obtain an extended grid. The regional features of the extended grid are compared with the boundary region features of the segmentation unit of the segmentation edge. When the boundary region features have a continuation relationship in the extended grid, it is determined that there is a cross-border defect. With the segmentation edge as the center, edge compensation segmentation is performed according to the corresponding scale segmentation target to obtain a compensated segmentation block and add it to the image grid segmentation library.

[0055] A glass fiber mesh defect detection system provided by an embodiment of the present invention can execute a glass fiber mesh defect detection method provided by any embodiment of the present invention, and has corresponding functional modules and beneficial effects of the execution method.

[0056] Although this application makes various references to certain modules in the system according to the embodiments of the present application, any number of different modules may be used and run on the user terminal and / or server. The various units and modules included are only divided according to functional logic, but are not limited to the above divisions, as long as the corresponding functions can be achieved. In addition, the specific names of the functional units are only for the purpose of distinguishing them from each other and are not intended to limit the scope of protection of the present invention.

[0057] The above specific embodiments do not constitute a limitation on the scope of protection of this application. Those skilled in the art should understand that various modifications, combinations, and substitutions may be made based on design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this application shall be included within the scope of protection of this application.

Claims

1. A method for detecting defects in a glass fiber mesh cloth, characterized in that: The method comprises: The target glass fiber mesh is moved and photographed by setting shooting control parameters to obtain detection target image information, wherein the shooting control parameters include shooting distance, shooting angle, and shooting parameters, wherein the shooting parameters meet preset shooting requirements; Acquire a placement background image, perform coupling feature analysis on the placement background image, and determine coupling influence parameters; Performing background decoupling on the detection target image information based on the coupling influence parameter to extract the target image; Performing grid segmentation on the target image to obtain an image grid segmentation library; traversing the image grid segmentation library using pre-extracted defect features to obtain defect detection results; The performing coupling feature analysis on the placement background image to determine coupling influence parameters includes: Extracting texture features, color distribution features, and illumination distribution features of the placed background image to obtain texture features, color distribution features, and illumination distribution features of the background image; Extracting texture features, color distribution features, and illumination distribution features of the target glass fiber mesh cloth to obtain texture features, color distribution features, and illumination distribution features of the background image of the target mesh cloth; According to the texture features, color distribution features, and illumination distribution features, respectively analyzing the image feature parameter differences and correlation influences between the background image features and the target mesh cloth, wherein the image feature parameter differences are features with more obvious differences, and the correlation influences are features with higher similarity; Obtaining background decoupled image features based on the image feature parameter differences, wherein the background decoupled image features are extracted based on the image feature parameter differences to extract image features caused only by the background; The coupling influence parameter is obtained according to the correlation influence of the texture features, color distribution features, and illumination distribution features. The coupling influence parameter refers to a specific parameter that describes the influence of background features on target image features.

2. The method for detecting defects in a glass fiber mesh cloth according to claim 1, wherein: Before performing mobile shooting of the target glass fiber mesh cloth by setting shooting control parameters, the method includes: Analyze the characteristics of the light source, the motion characteristics of the detection target, and the constraints of the shooting equipment settings in the detection environment to obtain the environmental acquisition characteristics; Extracting shooting influencing parameters according to the environmental acquisition features, performing compensation analysis based on the shooting influencing parameters, and obtaining compensation target parameters; Taking the compensation target parameters as setting targets, relevant compensation calculations are performed on the shooting distance, shooting angle, and shooting parameters of the shooting equipment, and the shooting parameters are obtained with the preset shooting requirements as constraints.

3. The method for detecting defects in a glass fiber mesh cloth according to claim 2, wherein: After obtaining the shooting parameters, the method further includes: Setting the device movement speed according to the target motion characteristics; Configure the shooting frame rate according to the moving speed of the device.

4. The method for detecting defects in a glass fiber mesh cloth according to claim 1, wherein: Performing background decoupling on the detection target image information based on the coupling influence parameter to extract the target image includes: Extracting key frames from the detected target image information, annotating the key frames according to the positioning coordinates, and splicing a unit image of the glass fiber mesh cloth based on the positioning coordinates, wherein the unit image of the glass fiber mesh cloth is image information corresponding to a unit identification structure having an edge boundary of the glass fiber mesh cloth; Perform background feature removal on the fiberglass mesh unit image according to the background decoupling image features to obtain initial target image information; constructing a coupling matrix based on the coupling influence parameters; The coupling matrix is ​​used to decouple the background parameters of the initial target image information, and the image feature parameters of the target grid cloth are restored to obtain the target image.

5. The method for detecting defects in a glass fiber mesh cloth according to claim 1, wherein: Performing grid segmentation on the target image to obtain an image grid segmentation library, including: Based on the defect type and the pre-extracted defect feature library, the proportional relationship between the defect distribution size and the mesh size is obtained, and a scale segmentation list of multiple defect types and defect levels is constructed. The scale segmentation list of multiple defect types and defect levels is used to guide the subsequent image segmentation process to ensure that defects of different types and levels can be accurately identified at the appropriate scale; Obtaining detection accuracy scale constraints; Matching and extracting the scale segmentation list according to the detection accuracy scale constraint to obtain a multi-scale segmentation target, wherein the multi-scale segmentation target refers to image segmentation operations performed within different scale ranges to meet the requirements of defect detection of different types and levels; Performing grid segmentation on the target image according to the multi-scale segmentation target to obtain a plurality of image grid segmentation libraries, each of which has a scale identifier; The scale identification is searched and matched with the scale segmentation list to obtain the corresponding defect type and defect level, and the defect type is identified for each image grid segmentation library.

6. The method for detecting defects in a glass fiber mesh cloth according to claim 5, wherein: The image grid segmentation library is traversed using the pre-extracted defect features to obtain defect detection results, including: Matching and extracting the defect type identification of the image grid segmentation library with the pre-extracted defect features to obtain target defect features; The target defect feature is used to traverse each segmentation grid feature of the image grid segmentation library, and when the target defect feature is successfully matched, the defect detection result is obtained.

7. The method for detecting defects in a glass fiber mesh cloth according to claim 5, wherein: Performing grid segmentation on the target image according to the multi-scale segmentation target, including: Performing grid segmentation on the target image according to the multi-scale segmentation target to obtain segmentation blocks and segmentation edges, wherein the segmentation blocks are stored as image grids; Expanding the boundary neighborhood of the segmentation edge to obtain an expanded grid; Comparing the regional features of the extended grid with the boundary regional features of the segmentation units of the segmentation edge line, and determining that a cross-border defect exists when the boundary regional features have a continuation relationship in the extended grid; Taking the segmentation edge as the center, edge compensation segmentation is performed according to the corresponding scale segmentation target, and the compensated segmentation block is obtained and added to the image grid segmentation library.

8. A defect detection system for glass fiber mesh cloth, characterized in that: The system is used to perform the defect detection method for glass fiber mesh cloth according to any one of claims 1 to 7, and the system comprises: An image acquisition module is used to perform mobile shooting of the target glass fiber mesh cloth by setting shooting control parameters to obtain detection target image information, wherein the shooting control parameters include shooting distance, shooting angle, and shooting parameters, wherein the shooting parameters meet preset shooting requirements; A coupling influence acquisition module is used to acquire a placement background image, perform coupling feature analysis on the placement background image, and determine coupling influence parameters; A background decoupling module is used to perform background decoupling on the detection target image information based on the coupling influence parameter to extract the target image; A grid segmentation module, configured to perform grid segmentation on the target image to obtain an image grid segmentation library; The detection result acquisition module is used to traverse the image grid segmentation library using pre-extracted defect features to obtain defect detection results.

Images