Cloth flaw detection system based on double-side multi-camera time-sharing stroboscopic imaging

Abstract

The invention discloses a cloth flaw detection system based on double-side multi-camera time-sharing stroboscopic imaging, and relates to the technical field of textile equipment. Comprising a multi-camera array used for collecting images of the surface of cloth, the multi-camera array comprises industrial camera sets arranged on the two sides of a rack of the cloth inspecting machine, the industrial camera sets are arranged in a staggered array mode in the cloth conveying direction, and the shooting ranges of the adjacent industrial cameras have overlapping areas; wherein the industrial camera group comprises six industrial cameras on both sides. According to the invention, through staggered array layout of six industrial cameras on each side and overlapped shooting ranges of the adjacent cameras, full-area coverage of the cloth surface is realized, and detection blind areas at edges and wrinkles are eliminated; the cloth conveying speed is collected by means of the transmission cooperation module, the main control unit dynamically adjusts the frame rate of the industrial camera and the frequency of the stroboscopic lamp according to the cloth conveying speed, a time-sharing stroboscopic mode is achieved in combination with the time schedule controller, it is ensured that clear images can still be obtained under high-speed conveying, and the problem of motion blur is solved.

Description

Technical Field

[0001] This invention relates to the field of textile equipment technology, specifically to a fabric defect detection system based on dual-side multi-camera time-division stroboscopic imaging. Background Technology

[0002] In the textile industry, defect detection on the fabric surface is a crucial step in ensuring product quality. After weaving and dyeing processes, fabrics may develop defects such as broken yarns, spot stains, and holes. These defects directly affect subsequent processing and the value of the finished product, thus requiring specialized fabric inspection equipment to examine the fabric surface. Currently, commonly used fabric inspection machines in the industry rely on cameras to capture images of the fabric surface, illuminated by a light source, and then transmit the images to an image processing unit. Defect recognition algorithms then determine whether the fabric has defects. With the increasing efficiency of textile production, most textile factories now transport fabric at speeds of 60-120 m / min, and the fabric width is generally 1.8-3 m. This places higher demands on the imaging coverage and clarity of fabric inspection machines. The cameras must ensure complete coverage of the entire fabric width, without missing edges, wrinkles, or other areas prone to concealing defects, while simultaneously capturing clear images at high speeds to provide a foundation for accurate defect identification by the algorithm.

[0003] However, in such high-speed, wide-width production scenarios, existing fabric inspection machines struggle to simultaneously achieve full-coverage imaging without blind spots and clear imaging without motion blur. Most existing fabric inspection machines employ a single camera or a layout of 3-4 cameras on one side. These cameras have limited shooting range and cannot fully cover the 1.8-3m width of the fabric, especially the edges and folds, which easily become blind spots due to the camera's lack of coverage, resulting in the inability to effectively capture minute defects in these areas. Furthermore, existing fabric inspection machines often use constant-light sources and fixed frame rates. When the fabric conveying speed reaches 60-120m / min, the images captured by the cameras under constant-light conditions experience significant motion blur due to the continuous movement of the fabric, causing the loss of texture details of defects and failing to meet the image clarity requirements of defect recognition algorithms. Even though some equipment attempts to expand coverage by increasing the number of cameras or improve imaging by increasing light source brightness, the timing of camera shooting and the timing of light emission are not designed in a coordinated manner. This not only makes it difficult to completely eliminate blind spots in the detection of fabric edges and folds, but also fails to effectively solve the motion blur problem under high-speed transmission. Furthermore, it is easily affected by ambient light such as fluorescent lights and natural light in the workshop, resulting in uneven brightness of the fabric surface image, further reducing the accuracy of defect identification. Ultimately, this leads to a high rate of missed and false detections in the fabric inspection process, making it unable to meet the needs of high-speed, wide-width fabric production and inspection. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides a fabric defect detection system based on dual-side multi-camera time-division stroboscopic imaging, which solves the problems of incomplete detection coverage, blurred imaging under high-speed transmission, and ambient light interference in existing technologies.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a fabric defect detection system based on dual-side multi-camera time-division stroboscopic imaging, comprising: A multi-camera array is used to acquire images of the fabric surface. The multi-camera array includes industrial camera groups set on both sides of the fabric inspection machine frame. The industrial camera groups are arranged in a staggered array in the fabric conveying direction, and the shooting ranges of adjacent industrial cameras have overlapping areas. The industrial camera group includes six industrial cameras on each side. A multi-light source system, which is set up in correspondence with a multi-camera array, is used to provide illumination to the fabric surface. The multi-light source system includes a strobe light group corresponding to each industrial camera in the multi-camera array. The light source wavelength of the strobe light group is located in a preset wavelength range, and each strobe light in the strobe light group has a preset flash duration. The strobe light is a narrow-band high-brightness LED strobe light. The transmission coordination module is used to collect the fabric conveying speed in real time and transmit the real-time collected conveying speed signal to the main control unit; wherein, the transmission coordination module is an encoder set at the fabric conveying roller shaft; The main control unit is connected to the multi-camera array, the multi-light source system and the transmission coordination module respectively. It is used to receive the transmission speed signal, dynamically adjust the frame rate of the industrial camera according to the transmission speed signal, and dynamically adjust the strobe frequency of the strobe light. The timing controller is connected to the multi-camera array and multi-light source system to control the multi-camera array and multi-light source system to work together to realize the time-division working mode of the stroboscopic lamp flashing synchronously when the industrial camera shutter is open and the stroboscopic lamp stopping flashing when the industrial camera shutter is closed. The image processing unit is connected to the multi-camera array to receive fabric images acquired by the multi-camera array, perform stitching processing on the fabric images, and perform noise reduction processing on the stitched images.

[0006] Furthermore, the industrial camera in the multi-camera array is a 4K or 8K industrial camera, which is used to acquire high-resolution images of the fabric surface; wherein, the preset angle formed between the lens axis of the industrial camera and the fabric surface is 15 degrees; wherein, the preset distance between the industrial camera and the fabric surface is 30 cm to 50 cm.

[0007] Furthermore, in the multi-camera array, the overlap rate of the overlapping areas of the shooting ranges of adjacent industrial cameras is 15%; wherein, the overlapping area is the overlapping part of the effective shooting range of each adjacent industrial camera, and the overlapping part is used to ensure continuous coverage of the fabric.

[0008] Furthermore, the light source wavelength of the strobe lamp in the multi-light source system is located in a preset wavelength range of 550 nm to 600 nm; wherein, the light source wavelength of the narrow-band high-brightness LED strobe lamp is located in the preset wavelength range; and the preset flash duration of the strobe lamp is ≤10 μs.

[0009] Furthermore, the transmission coordination module is an encoder installed on the fabric conveying roller shaft; the encoder is used to accurately and in real time collect the fabric conveying speed; wherein, the fabric conveying speed ranges from 30 m / min to 120 m / min, and the encoder transmits the accurately collected conveying speed signal to the main control unit.

[0010] Furthermore, the main control unit dynamically adjusts the frame rate of each industrial camera in the multi-camera array; the frame rate adjustment range of the industrial cameras is 50 frames per second to 200 frames per second; the main control unit also dynamically adjusts the strobe frequency of each strobe lamp in the multi-light source system; the main control unit ensures that the number of images captured per meter of fabric is stable and not less than 30 through dynamic adjustment.

[0011] Furthermore, the industrial cameras in the multi-camera array are equipped with narrowband filter lenses; wherein, the narrowband filter lenses are used to work in conjunction with the multi-light source system; the narrowband filter lenses only allow light within a preset wavelength range to pass through, and are used to filter ambient light.

[0012] Furthermore, after receiving the fabric images acquired by the multi-camera array, the image processing unit performs stitching processing on the fabric images; The image processing unit then performs noise reduction processing on the stitched fabric image; The image processing unit uploads the noise-reduced fabric image to the cloud system for defect identification.

[0013] Furthermore, the working process of the fabric defect detection system also includes: After the system starts, the transmission coordination module collects the fabric conveying speed in real time and transmits the conveying speed signal to the main control unit. The main control unit calculates and sets the frame rate of each industrial camera in the multi-camera array and the flash frequency of each strobe lamp in the multi-light source system based on the received transmission speed signal. When the fabric begins to be conveyed on the conveyor roller, the multi-camera array opens the shutter to capture images according to the set frame rate. The timing controller synchronously controls the strobe lights and shutter opening in the multi-light source system to perform strobe flashing; The fabric images captured by the multi-camera array are transmitted in real time to the image processing unit for processing.

[0014] Furthermore, the cloud system is connected to a defect detection algorithm module; The defect detection algorithm module is used to perform in-depth analysis on the noise-reduced fabric images uploaded by the image processing unit; The in-depth analysis includes identifying features such as broken yarns, spot stains, and holes in fabric images.

[0015] Compared with the prior art, the beneficial effects of the present invention are as follows: This invention achieves full coverage of the fabric surface by using a staggered array of six industrial cameras on each side, with overlapping shooting ranges of adjacent cameras, eliminating blind spots at edges and folds. A transmission coordination module collects the fabric conveying speed, and the main control unit dynamically adjusts the frame rate of the industrial cameras and the frequency of the strobe lights accordingly. Combined with a time-division strobe mode implemented by a timing controller, clear images are still obtained even at high speeds, solving the motion blur problem. Simultaneously, a narrow-band high-brightness LED strobe light, in conjunction with the narrow-band filter lens of the industrial camera, effectively filters ambient light, improves image brightness uniformity, and reduces interference from ambient light. Through the coordinated work of all modules, the overall invention enhances the comprehensiveness, clarity, and accuracy of fabric defect detection, adapting to the detection needs of high-speed, wide-width fabric production. Attached Figure Description

[0016] Figure 1 This is a system structure diagram of the present invention. Detailed Implementation

[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] Please see Figure 1 This invention provides a fabric defect detection system based on dual-side multi-camera time-division stroboscopic imaging, comprising: A multi-camera array is used to acquire images of the fabric surface. The multi-camera array includes industrial camera groups set on both sides of the fabric inspection machine frame. The industrial camera groups are arranged in a staggered array in the fabric conveying direction, and the shooting ranges of adjacent industrial cameras have overlapping areas. The industrial camera group includes six industrial cameras on each side. A multi-light source system, which is set up in correspondence with a multi-camera array, is used to provide illumination to the fabric surface. The multi-light source system includes a strobe light group corresponding to each industrial camera in the multi-camera array. The light source wavelength of the strobe light group is located in a preset wavelength range, and each strobe light in the strobe light group has a preset flash duration. The strobe light is a narrow-band high-brightness LED strobe light. The transmission coordination module is used to collect the fabric conveying speed in real time and transmit the real-time collected conveying speed signal to the main control unit; wherein, the transmission coordination module is an encoder set at the fabric conveying roller shaft; The main control unit is connected to the multi-camera array, the multi-light source system and the transmission coordination module respectively. It is used to receive the transmission speed signal, dynamically adjust the frame rate of the industrial camera according to the transmission speed signal, and dynamically adjust the strobe frequency of the strobe light. The timing controller is connected to the multi-camera array and multi-light source system to control the multi-camera array and multi-light source system to work together to realize the time-division working mode of the stroboscopic lamp flashing synchronously when the industrial camera shutter is open and the stroboscopic lamp stopping flashing when the industrial camera shutter is closed. The image processing unit is connected to the multi-camera array to receive fabric images acquired by the multi-camera array, perform stitching processing on the fabric images, and perform noise reduction processing on the stitched images.

[0019] Specifically, when building a fabric defect detection system based on dual-sided multi-camera time-division stroboscopic imaging, the first step is to construct a multi-camera array. This array needs to be set on both sides of the fabric inspection machine frame, with six industrial cameras installed on each side to form an industrial camera group. The industrial camera groups adopt a staggered array layout in the fabric conveying direction, while ensuring that the shooting range of adjacent industrial cameras overlaps. This layout can effectively cover the fabric surface, especially the fabric edges and wrinkles that are difficult to cover by traditional fabric inspection machines, avoiding the occurrence of blind spots in detection.

[0020] Next, a multi-light source system is configured. This system must correspond one-to-one with the multi-camera array. Each industrial camera is paired with a set of strobe lights to form a strobe light group. Narrow-band high-brightness LED strobe lights are selected as the light source, and the wavelength of the strobe lights is controlled within a preset range. At the same time, the flash duration of each strobe light is set to a preset value. The characteristics of narrow-band light source help to highlight the features of fabric defects, and short flash duration can reduce the impact of fabric movement on imaging, improving the motion blur problem that easily occurs in images under traditional constant light sources.

[0021] Then, the transmission coordination module is installed, and the encoder is set at the fabric conveyor roller. The encoder can collect the fabric conveying speed in real time and accurately transmit the collected speed signal to the main control unit, providing a precise basis for subsequent parameter adjustment and solving the problem that traditional equipment is difficult to adapt to different fabric conveying speeds.

[0022] The main control unit needs to establish connections with the multi-camera array, the multi-light source system, and the drive coordination module. Its core function is to dynamically adjust the frame rate of the industrial camera and the flash frequency of the strobe lamp based on the transmission speed signal transmitted by the drive coordination module. A frame rate calculation model is used here, and the formula is: ; In the formula, This refers to the frame rate of the industrial camera, with a value ranging from 50 to 200 frames per second. The number of images taken per meter of fabric must meet the following requirements. To ensure detection accuracy; The fabric conveying speed is collected in real time by an encoder. The flash frequency of the strobe light must be consistent with the frame rate of the industrial camera, i.e., the flash frequency equals... This is to ensure the synchronization of the light source and the camera shutter.

[0023] The timing controller connects the multi-camera array and the multi-light source system, enabling them to work together through timing control. When the industrial camera shutter opens, the timing controller controls the strobe lights to flash synchronously; when the industrial camera shutter closes, the strobe lights stop flashing, forming a time-sharing working mode. This mode can further reduce motion blur, improve image sharpness, and avoid unnecessary energy consumption of the light source.

[0024] The image processing unit is connected to a multi-camera array. After receiving fabric images acquired by the array, it first stitches the images together, integrating the scattered images from different cameras into a complete fabric surface image. Then, it performs noise reduction processing on the stitched image to remove interference signals and improve image quality. The entire system, through the collaborative work of its modules, effectively solves the problems of incomplete coverage and blurry imaging in traditional fabric inspection machines, thus improving the reliability of fabric defect detection.

[0025] Specifically, in a multi-camera array, 4K or 8K resolution industrial cameras are selected. Their high resolution allows them to capture richer details of the fabric surface, facilitating the identification of minute defects and resolving the issue of lost detail due to insufficient resolution in traditional cameras. When installing the industrial camera, the angle between the camera lens axis and the fabric surface should be adjusted to a 15-degree angle. This angle effectively avoids reflections caused by direct sunlight on the fabric surface, reducing interference with image quality. Simultaneously, the distance between the industrial camera and the fabric surface should be controlled between 30 and 50 centimeters. This distance range ensures the camera has sufficient coverage of the fabric area while maintaining image clarity even at greater distances, further enhancing the image's detail rendering capabilities and providing a high-quality image foundation for defect identification.

[0026] In this embodiment, the overlap rate of the overlapping areas of the shooting ranges of adjacent industrial cameras in the multi-camera array is 15%; wherein, the overlapping area is the overlapping part of the effective shooting range of each adjacent industrial camera, and the overlapping part is used to ensure continuous coverage of the fabric.

[0027] Specifically, when deploying industrial cameras in a multi-camera array, it is crucial to strictly control the overlapping area of ​​the shooting ranges of adjacent industrial cameras, ensuring an overlap rate of 15%. The effective shooting ranges of adjacent industrial cameras will form an overlapping portion, which fills any possible shooting gaps between adjacent cameras. Regardless of the fabric's position during transport, whether it's the middle or edge area, it can be continuously covered by the cameras. This eliminates the blind spots that occur in traditional equipment due to non-overlapping shooting ranges, ensuring that every area of ​​the fabric surface can be accurately imaged, thus guaranteeing comprehensive detection of fabric defects.

[0028] In this embodiment, the light source wavelength of the strobe lamp in the multi-light source system is within a preset wavelength range of 550 nm to 600 nm; wherein, the light source wavelength of the narrow-band high-brightness LED strobe lamp is within the preset wavelength range; and the preset flash duration of the strobe lamp is ≤10 μs.

[0029] Specifically, in a multi-light source system, the wavelength of the strobe light is set within a preset range of 550 to 600 nanometers. When light within this wavelength range illuminates the fabric surface, it creates a more pronounced contrast between the fabric's texture and imperfections, facilitating the identification of defects in subsequent images. Simultaneously, a narrow-band, high-brightness LED strobe light is selected, with its flash duration controlled to no more than 10 microseconds. This extremely short flash duration allows for instantaneous image capture of the fabric surface during high-speed transport, minimizing image blurring and motion blur caused by fabric movement. This solves the problem of image blurring in high-speed scenarios using traditional constant-brightness light sources, improving image clarity during high-speed transport.

[0030] In this embodiment, the transmission coordination module is an encoder installed on the fabric conveying roller shaft; the encoder is used to accurately and in real time collect the fabric conveying speed; wherein, the fabric conveying speed ranges from 30 m / min to 120 m / min, and the encoder transmits the accurately collected conveying speed signal to the main control unit.

[0031] Specifically, the transmission coordination module uses an encoder, which is installed on the fabric conveyor roller. The encoder directly contacts the conveyor roller to accurately and in real time acquire the fabric conveying speed. Regardless of the fabric conveying speed range of 30 m / min to 120 m / min, the encoder can accurately capture speed changes. The acquired conveying speed signal is transmitted to the main control unit in real time, providing the main control unit with accurate speed data to adjust the industrial camera frame rate and strobe light flash frequency. This ensures that the system can adjust its working parameters in a timely manner according to the actual fabric conveying speed, adapting to different production speed requirements and avoiding the problem of decreased detection effectiveness when speed changes in traditional equipment due to fixed speed parameters.

[0032] In this embodiment, the main control unit dynamically adjusts the frame rate of each industrial camera in the multi-camera array; the frame rate adjustment range of the industrial cameras is 50 frames per second to 200 frames per second; the main control unit also dynamically adjusts the strobe frequency of each strobe lamp in the multi-light source system; the main control unit ensures that the number of images captured per meter of fabric is stable and not less than 30 through dynamic adjustment.

[0033] Specifically, after receiving the fabric conveying speed signal from the transmission coordination module, the main control unit dynamically adjusts the frame rate of each industrial camera in the multi-camera array and the flash frequency of each strobe light in the multi-light source system based on the speed value. The frame rate adjustment range is controlled between 50 frames per second and 200 frames per second. The flash frequency of the strobe lights must be consistent with the frame rate of the industrial cameras to ensure synchronization between the light source and the camera shutter. Through this dynamic adjustment, regardless of whether the fabric conveying speed increases or decreases, the number of images captured per meter of fabric remains stable and is no less than 30. This ensures that each area of ​​the fabric surface has sufficient image data for inspection, avoiding missed defects due to insufficient image quantity and improving the comprehensiveness and reliability of the inspection.

[0034] In this embodiment, the industrial camera in the multi-camera array is equipped with a narrowband filter lens; wherein, the narrowband filter lens is used to work in conjunction with the multi-light source system; the narrowband filter lens only allows light within a preset wavelength range to pass through, and is used to filter ambient light.

[0035] Specifically, each industrial camera in the multi-camera array is equipped with a narrowband filter lens. This narrowband filter lens must be compatible with the wavelength of the strobe light source in the multi-light source system, allowing only light within a preset wavelength range to pass through. Ambient light in the workshop, such as fluorescent lights and natural light, is filtered out by the narrowband filter lens because its wavelength is outside the preset range. This effectively reduces the interference of ambient light on fabric image acquisition, avoids uneven brightness in the fabric surface image caused by ambient light, and makes the acquired image brightness and contrast more stable, resulting in better image quality. This provides more accurate image data for subsequent image processing and defect identification, reducing the impact of ambient light on the inspection results.

[0036] In this embodiment, after receiving the fabric images acquired by the multi-camera array, the image processing unit performs stitching processing on the fabric images. The image processing unit then performs noise reduction processing on the stitched fabric image; The image processing unit uploads the noise-reduced fabric image to the cloud system for defect identification.

[0037] Specifically, after receiving the fabric images acquired by the multi-camera array, the image processing unit first performs a stitching process. Since the multi-camera array consists of multiple cameras on both sides, the acquired images are scattered local images. The stitching process integrates these local images according to the actual location of the fabric into a complete image of the fabric surface, ensuring that no image areas are missed. After stitching, the image processing unit performs noise reduction processing on the stitched image to remove noise and interference signals caused by environmental interference and equipment noise, making the image cleaner and clearer. The noise-reduced fabric image is then uploaded to the cloud system, providing high-quality image data for subsequent defect identification. Through stitching and noise reduction processing, the integrity and clarity of the image are improved, laying the foundation for accurate defect identification.

[0038] In this embodiment, the operation of the fabric defect detection system further includes: After the system starts, the transmission coordination module collects the fabric conveying speed in real time and transmits the conveying speed signal to the main control unit. The main control unit calculates and sets the frame rate of each industrial camera in the multi-camera array and the flash frequency of each strobe lamp in the multi-light source system based on the received transmission speed signal. When the fabric begins to be conveyed on the conveyor roller, the multi-camera array opens the shutter to capture images according to the set frame rate. The timing controller synchronously controls the strobe lights and shutter opening in the multi-light source system to perform strobe flashing; The fabric images captured by the multi-camera array are transmitted in real time to the image processing unit for processing.

[0039] Specifically, after the fabric defect detection system is activated, the encoder of the transmission coordination module immediately begins to work, acquiring the fabric conveying speed in real time and continuously transmitting the acquired conveying speed signal to the main control unit. Upon receiving the speed signal, the main control unit calculates the speed based on the frame rate model. The system calculates the frame rate of each industrial camera in the multi-camera array and the flash frequency of each strobe light in the multi-light source system at the current speed, and sets the calculated parameters to the corresponding industrial cameras and strobe lights. When the fabric begins to move on the conveyor rollers, the multi-camera array opens its shutter at the set frame rate to capture images of the fabric surface. Simultaneously, the timing controller synchronously controls the strobe lights in the multi-light source system to flash at the instant the industrial camera shutter opens, providing sufficient and appropriate illumination for image acquisition. The fabric images captured by the multi-camera array are transmitted to the image processing unit in real time. The image processing unit performs subsequent image processing such as stitching and noise reduction according to a preset procedure. The entire process is coherent and orderly, with each module working in concert to ensure the timeliness and accuracy of image acquisition and processing, adapting to the continuous fabric conveying and inspection requirements.

[0040] In this embodiment, the cloud system is connected to a defect detection algorithm module; The defect detection algorithm module is used to perform in-depth analysis on the noise-reduced fabric images uploaded by the image processing unit; The in-depth analysis includes identifying features such as broken yarns, spot stains, and holes in fabric images.

[0041] Specifically, the cloud system is connected to a dedicated defect detection algorithm module. After the image processing unit uploads the noise-reduced fabric images to the cloud system, the defect detection algorithm module performs in-depth analysis on these images using a weighted feature fusion model to identify defects such as broken yarns, spot stains, and holes. The algorithm formula is as follows: In the formula, The defect judgment value is when It was determined to be defective at that time. The preset judgment threshold is determined through sample training; These are the weighting coefficients for grayscale difference, texture complexity, and shape factor, respectively, satisfying... The values ​​are determined using the Analytic Hierarchy Process (AHP). The specific application steps are as follows: First, construct a judgment matrix, using grayscale difference, texture complexity, and shape factor as criteria layer elements. Determine the relative importance of each element through pairwise comparison, assigning values ​​using a 1-9 scale: 1 indicates equal importance, 3 indicates slightly important, 5 indicates significantly important, 7 indicates strongly important, and 9 indicates extremely important. 2, 4, 6, and 8 are intermediate values. Next, calculate the largest eigenvalue of the judgment matrix and its corresponding eigenvector. Normalize the eigenvectors to obtain the weight coefficients. Finally, a consistency check is performed, and the consistency index CI = (λmax - n) / (n - 1) is calculated, where λmax is the largest eigenvalue and n is the number of elements in the criterion layer. The consistency index CI is compared with the random consistency index RI. When CR = CI / RI < 0.1, the judgment matrix meets the consistency requirements and the weight coefficients are valid. The normalized difference in grayscale between the defect and the background is calculated as follows: ,in This represents the actual grayscale value, ranging from 0 to 255. These are the minimum and maximum values ​​of the grayscale difference in the sample set, respectively; The normalized texture complexity is calculated as follows: ,in To calculate the texture entropy using the gray-level co-occurrence matrix (GLCM), the parameters of the GLCM are set as follows: distance d = 1 pixel, angle θ = 0°, 45°, 90°, 135°, taking the average of the four directions to avoid directional bias, and 256 gray levels. These are the minimum and maximum values ​​of the texture entropy in the sample set, respectively; The normalized shape factor is calculated as follows: ,in The roundness of the defective area is defined as: A represents the area of ​​the defective region, and L represents the perimeter of the defective region, taking values ​​from 0 to 1. For a circular shape, S=1; the more irregular the shape, the closer S is to 0. These represent the minimum and maximum values ​​of circularity in the sample set, respectively. This algorithm avoids the influence of differences in feature dimensions on the judgment results by weighted fusion of normalized multi-features. It can accurately identify broken yarns, dotted stains, and holes in fabric images, avoiding errors caused by subjectivity and fatigue in manual identification, thus improving the efficiency and accuracy of defect identification and providing a basis for subsequent fabric quality judgment and processing.

[0042] In summary, this invention achieves full coverage of the fabric surface by using a staggered array of six industrial cameras on each side, with overlapping shooting ranges of adjacent cameras, eliminating blind spots at edges and folds. The transmission coordination module collects the fabric conveying speed, and the main control unit dynamically adjusts the frame rate of the industrial cameras and the frequency of the strobe lamp accordingly. Combined with a time-division strobe mode implemented by a timing controller, clear images are still obtained even at high speeds, solving the motion blur problem. Simultaneously, a narrow-band high-brightness LED strobe lamp, in conjunction with the narrow-band filter lens of the industrial camera, effectively filters ambient light, improves image brightness uniformity, and reduces interference from ambient light. Overall, through the coordinated work of all modules, the invention improves the comprehensiveness, clarity, and accuracy of fabric defect detection, adapting to the detection needs of high-speed, wide-width fabric production.

[0043] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0044] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A fabric defect detection system based on dual-side multi-camera time-division stroboscopic imaging, characterized in that, include: A multi-camera array is used to acquire images of the fabric surface. The multi-camera array includes industrial camera groups set on both sides of the fabric inspection machine frame. The industrial camera groups are arranged in a staggered array in the fabric conveying direction, and the shooting ranges of adjacent industrial cameras have overlapping areas. The industrial camera group includes six industrial cameras on each side. A multi-light source system, which is set up in correspondence with a multi-camera array, is used to provide illumination to the fabric surface. The multi-light source system includes a strobe light group corresponding to each industrial camera in the multi-camera array. The light source wavelength of the strobe light group is located in a preset wavelength range, and each strobe light in the strobe light group has a preset flash duration. The strobe light is a narrow-band high-brightness LED strobe light. The transmission coordination module is used to collect the fabric conveying speed in real time and transmit the real-time collected conveying speed signal to the main control unit; wherein, the transmission coordination module is an encoder set at the fabric conveying roller shaft; The main control unit is connected to the multi-camera array, the multi-light source system and the transmission coordination module respectively. It is used to receive the transmission speed signal, dynamically adjust the frame rate of the industrial camera according to the transmission speed signal, and dynamically adjust the strobe frequency of the strobe light. The timing controller is connected to the multi-camera array and multi-light source system to control the multi-camera array and multi-light source system to work together, so as to realize the time-division working mode of the stroboscopic lamp flashing synchronously when the industrial camera shutter is open and the stroboscopic lamp stopping flashing when the industrial camera shutter is closed. The image processing unit is connected to the multi-camera array to receive fabric images acquired by the multi-camera array, perform stitching processing on the fabric images, and perform noise reduction processing on the stitched images.

2. The fabric defect detection system based on dual-side multi-camera time-division stroboscopic imaging according to claim 1, characterized in that, The industrial cameras in the multi-camera array are 4K or 8K industrial cameras used to acquire high-resolution images of the fabric surface; wherein the preset angle between the lens axis of the industrial camera and the fabric surface is 15 degrees; wherein the preset distance between the industrial camera and the fabric surface is 30 cm to 50 cm.

3. The fabric defect detection system based on dual-side multi-camera time-division stroboscopic imaging according to claim 1, characterized in that, In the multi-camera array, the overlap rate of the overlapping areas of the shooting ranges of adjacent industrial cameras is 15%; wherein, the overlapping area is the overlapping part of the effective shooting range of each adjacent industrial camera, and the overlapping part is used to ensure continuous coverage of the fabric.

4. The fabric defect detection system based on dual-side multi-camera time-division stroboscopic imaging according to claim 1, characterized in that, The strobe light in the multi-light source system has a light source wavelength in the preset wavelength range of 550 nm to 600 nm; wherein, the light source wavelength of the narrow-band high-brightness LED strobe light is in the preset wavelength range; and the preset flash duration of the strobe light is ≤10 μs.

5. The fabric defect detection system based on dual-side multi-camera time-division stroboscopic imaging according to claim 1, characterized in that, The transmission coordination module is an encoder installed on the fabric conveying roller shaft; the encoder is used to accurately and in real time collect the fabric conveying speed; wherein, the fabric conveying speed ranges from 30 m / min to 120 m / min, and the encoder transmits the accurately collected conveying speed signal to the main control unit.

6. The fabric defect detection system based on dual-side multi-camera time-division stroboscopic imaging according to claim 1, characterized in that, The main control unit dynamically adjusts the frame rate of each industrial camera in the multi-camera array; the frame rate adjustment range of the industrial cameras is 50 frames per second to 200 frames per second; the main control unit also dynamically adjusts the strobe frequency of each strobe lamp in the multi-light source system; the main control unit ensures that the number of images captured per meter of fabric is stable and not less than 30 through dynamic adjustment.

7. The fabric defect detection system based on dual-side multi-camera time-division stroboscopic imaging according to claim 1, characterized in that, The industrial cameras in the multi-camera array are equipped with narrowband filter lenses; the narrowband filter lenses are used to work in conjunction with the multi-light source system; the narrowband filter lenses only allow light within a preset wavelength range to pass through, and are used to filter ambient light.

8. The fabric defect detection system based on dual-side multi-camera time-division stroboscopic imaging according to claim 1, characterized in that, After receiving the fabric images acquired by the multi-camera array, the image processing unit performs stitching processing on the fabric images. The image processing unit then performs noise reduction processing on the stitched fabric image; The image processing unit uploads the noise-reduced fabric image to the cloud system for defect identification.

9. The fabric defect detection system based on dual-side multi-camera time-division stroboscopic imaging according to claim 1, characterized in that, The working process of the fabric defect detection system also includes: After the system starts, the transmission coordination module collects the fabric conveying speed in real time and transmits the conveying speed signal to the main control unit. The main control unit calculates and sets the frame rate of each industrial camera in the multi-camera array and the flash frequency of each strobe lamp in the multi-light source system based on the received transmission speed signal. When the fabric begins to be conveyed on the conveyor roller, the multi-camera array opens the shutter to capture images according to the set frame rate. The timing controller synchronously controls the strobe lights and shutter opening in the multi-light source system to perform strobe flashing; The fabric images captured by the multi-camera array are transmitted in real time to the image processing unit for processing.

10. The fabric defect detection system based on dual-side multi-camera time-division stroboscopic imaging according to claim 8, characterized in that, The cloud system is connected to a defect detection algorithm module; The defect detection algorithm module is used to perform in-depth analysis on the noise-reduced fabric images uploaded by the image processing unit; The in-depth analysis includes identifying features such as broken yarns, spot stains, and holes in fabric images.

Images