Fabric Defect Detection Methods and Systems

Abstract

This application relates to a method and system for detecting fabric defects, specifically in the field of fabric defect detection. The method includes: receiving several actual images of fabric areas from an image acquisition device group, along with the identification codes of the image acquisition devices corresponding to the target actual images; retrieving a standard image corresponding to the identification code from a preset fabric image reference database; if the actual similarity value obtained by comparing the actual image and the standard image is less than a preset similarity standard value, then determining that the fabric area in the actual image has a defect; and sending a warning message related to the presence of a defect in the fabric area of ​​the actual image to the user terminal. The technical advantage of this application is that it automatically detects fabric defects during the fabric production process, resulting in high detection efficiency.

Description

Technical Field

[0001] This invention relates to the technical field of fabric surface defect detection, and in particular to a method and system for fabric surface defect detection. Background Technology

[0002] If a malfunction occurs during production on a double-needle bed warp knitting machine, the produced fabric will be defective.

[0003] In related technologies, the detection of defects in fabrics is mostly done manually. Manual defect detection is labor-intensive, inefficient, and prone to missed or incorrect detections, which leads to an increase in the number of defective fabrics. Summary of the Invention

[0004] To address the problem of increased fabric defect rates due to low efficiency in manual defect detection, this application provides a method and system for detecting fabric defects.

[0005] Firstly, this application provides a method for detecting fabric defects. This method is applied to a double-needle bed warp knitting machine, which is equipped with an image acquisition device group. The image acquisition device group is used to capture images of the front or back of the fabric. The method employs the following technical solution:

[0006] Receive image information sent by the image acquisition device group, the image information including several actual images with cloth areas and the identification code of the image acquisition device corresponding to the target actual image;

[0007] Retrieve the standard image corresponding to the target's identity code from the preset fabric image reference database;

[0008] The actual image is compared with the standard image to obtain the actual similarity value;

[0009] If the actual similarity value is less than the preset similarity standard value, it is determined that there are defects in the fabric area of ​​the actual image.

[0010] Send a warning message to the user terminal regarding the presence of defects in the fabric area shown in the actual image.

[0011] By adopting the above technical solution, the double-needle bed warp knitting machine uses an image acquisition device group to capture real-time images of the front or back of the fabric during the fabric production process. By retrieving the standard image corresponding to the identification code of the image acquisition device corresponding to the actual image from a preset fabric image reference database, if the actual similarity value obtained by comparing the actual image and the standard image is less than the preset similarity standard value, it indicates that there is a defect in the fabric area in the actual image. This realizes automatic detection of fabric defects, replacing manual inspection, saving time and effort. Furthermore, by using multiple image acquisition devices to capture images of the front or back of the fabric, it helps to avoid blind spots and missed detections, resulting in good defect detection effect.

[0012] After determining that there are defects in the fabric area in the actual image, an early warning message related to the defects in the fabric area in the actual image is sent to the user's terminal so that the user can be informed of the defects in the fabric in a timely manner. Then, the user can repair the double needle bed warp knitting machine according to the defects in the fabric, so as to avoid the double needle bed warp knitting machine from continuously producing fabrics with defects, which helps to reduce the amount of defective fabrics.

[0013] Optionally, before comparing the actual image with the standard image, the method further includes:

[0014] According to the preset image cropping rules, the actual image is cropped to obtain the corrected image;

[0015] Set the corrected image as the actual image.

[0016] By adopting the above technical solution, before comparing the actual image with the standard image, a correction image cropped from the actual image is set as the actual image. Therefore, the image compared with the standard image is actually the correction image. Since the image captured by the image acquisition device may contain the equipment structure of the double needle bed warp knitting machine, cropping a portion of the actual image helps reduce the probability of other impurities in the final actual image compared with the standard image. This helps improve the accuracy of the similarity comparison results between the final actual image and the standard image, minimizes the probability of misjudgment, and improves the accuracy of defect detection results.

[0017] Optionally, determining that there are defects in the fabric area of ​​the actual image includes:

[0018] Based on the preset flaw outline range extraction rules, extract the actual flaw outline region corresponding to the difference between the actual image and the standard image in the actual image;

[0019] Identify the actual defect features of the actual defect outline region, the actual defect features including actual defect color features and actual defect texture features;

[0020] The system queries a preset defect category database to determine if the actual defect feature exists. If it does, it determines that the fabric area in the actual image has a defect. If not, it retrieves a preset obstacle feature database.

[0021] The system queries a preset obstacle feature database to determine if the actual defect feature exists. If it does, it determines that there is an obstacle in the fabric area of ​​the actual image. If not, it determines that the lens surface of the image acquisition device is dirty.

[0022] By employing the above technical solution, the actual defect contour regions corresponding to the differences between the actual image and the standard image are extracted, and the actual defect features of the actual defect contour regions are identified. By determining whether the actual defect features are located in the preset defect category data, if so, it indicates that there are defects in the fabric area of ​​the actual image; if not, it indicates that there are obstacles in the fabric area of ​​the actual image or that there is dirt on the surface of the lens of the image acquisition device, resulting in the calculated actual similarity value being less than the preset similarity standard value. By adopting the above solution, the authenticity of the actual defect features as fabric defect features is verified, and the phenomenon of misjudgment is avoided as much as possible, thereby helping to improve the accuracy of determining whether there are defects in the fabric.

[0023] Optionally, after determining that there is an obstacle in the fabric area of ​​the actual image, the method further includes:

[0024] Send a notification message to the user's device that indicates an obstacle exists in the fabric area shown in the actual image;

[0025] If, within a preset first time period, a confirmation message is received from the user terminal that the obstacle in the actual image is the equipment structure of a double-needle bed warp knitting machine, then the actual image will be updated to a standard image.

[0026] By adopting the above technical solution, if an obstacle is detected in the fabric area of ​​the actual image, a reminder message related to the obstacle is sent to the user's terminal so that the user can be informed in a timely manner. The user can then confirm whether the obstacle in the actual image is the structure of a double-needle bed warp knitting machine. If the obstacle is indeed the structure of a double-needle bed warp knitting machine, the user can send confirmation information through the terminal, further improving the verification that the fabric area in the actual image is free of defects. At this point, the actual image can be updated to a standard image, which contains the obstacle in the fabric area. This helps improve the accuracy of image similarity comparison and achieves deep learning of the standard images in the fabric image reference database, thus improving the accuracy of defect detection results.

[0027] Optionally, sending warning information related to defects in the fabric area in the actual image to the user terminal includes:

[0028] Retrieve the defect name corresponding to the actual defect characteristics from the preset defect type database;

[0029] If the defect name is found in the preset database of allowed early warning defects, the double needle bed warp knitting machine is stopped and an early warning message related to the defect in the fabric area in the actual picture is sent to the user terminal.

[0030] By adopting the above technical solution, after determining that there are defects in the fabric area of ​​the actual image, the defect name corresponding to the actual defect characteristics is retrieved from the preset defect type database. If the defect name exists in the preset allowed warning defect database, it means that the fabric defect name in the actual image is a defect that affects the fabric quality. Therefore, the double-needle bed warp knitting machine is stopped, which helps avoid the production of a large number of defective products due to the double-needle bed warp knitting machine continuing to operate, thus helping to reduce the amount of defective fabric. Simultaneously, warning information related to the presence of defects in the fabric area of ​​the actual image is sent to the user terminal to promptly remind the user of the defect. The user can then repair the double-needle bed warp knitting machine according to the defect situation, minimizing the production of defective fabric. The above-mentioned method of determining whether the defect name is an allowed warning defect name helps avoid the phenomenon where a defect name corresponding to a defect on the fabric does not affect the fabric quality but triggers a warning.

[0031] Optionally, determining that the lens surface of the image acquisition device is dirty includes:

[0032] Identify the actual sharpness of the actual image;

[0033] Determine whether the actual sharpness is within the preset sharpness standard range; if not, determine that the lens surface of the image acquisition device is dirty.

[0034] Supplement the air source to the lens surface of the image acquisition device described in the target;

[0035] After a preset second time period, a cloth shooting instruction is sent to the image acquisition device corresponding to the identity code;

[0036] Obtain the updated image returned by the image acquisition device corresponding to the identity code;

[0037] Identify the updated sharpness of the updated image;

[0038] If the updated sharpness is within the preset sharpness standard range, it is determined that the lens surface of the image acquisition device is dirty.

[0039] By adopting the above technical solution, after determining that the actual clarity of the actual image is outside the preset clarity standard range, the surface of the lens is cleaned by supplementing the air source. This helps to avoid affecting the clarity of the cloth image captured by the subsequent image acquisition equipment, and helps to prevent the actual image from having dirt on the lens surface of the image acquisition equipment, which would result in low similarity between the actual image and the standard image.

[0040] Optionally, the image acquisition device group includes several front image acquisition devices for capturing images of the front side of the fabric and several back image acquisition devices for capturing images of the back side of the fabric. The double-needle bed warp knitting machine is equipped with several back-light devices corresponding to the front image acquisition devices and several front back-light devices corresponding to the back image acquisition devices. Before receiving the image information sent by the image acquisition device group, the following is also included:

[0041] The backlight device that controls the supplementary light source on the reverse side of the fabric to start the supplementary lighting operation, and the backlight device that controls the supplementary light source on the front side of the fabric to stop the supplementary lighting operation;

[0042] Send a frontal shooting command to a plurality of the aforementioned frontal image acquisition devices, wherein the frontal shooting command includes a frontal identification code corresponding to the target frontal image acquisition device;

[0043] The backlight device that provides supplementary light source to the reverse side of the fabric stops its supplementary lighting operation, and the backlight device that provides supplementary light source to the front side of the fabric starts its supplementary lighting operation.

[0044] Send a reverse image capture command to a plurality of the reverse image acquisition devices, wherein the reverse image capture command includes a reverse identity code corresponding to the target reverse image acquisition device.

[0045] By adopting the above technical solution, when it is necessary to control the front image acquisition device to take pictures of the front of the fabric, the backlight device that provides supplementary light to the back of the fabric is controlled to start supplementary lighting, while the backlight device that provides supplementary light to the front of the fabric is controlled to stop supplementary lighting. This prevents the front image acquisition device from shooting with backlight, which helps to avoid exposure problems that would result in unclear fabric images. Ultimately, when taking pictures of both sides of the fabric, the backlight devices that provide supplementary light to the front and back of the fabric are controlled to operate alternately, which helps to avoid exposure problems in the processed fabric images. Furthermore, by illuminating the fabric, it helps to avoid dim lighting that would result in unclear images and affect the accuracy of the similarity comparison between the actual image and the standard image, thus improving the accuracy of fabric defect detection.

[0046] Secondly, this application provides a fabric defect detection system. This system is applied to a double-needle bed warp knitting machine, which is equipped with an image acquisition device group. The image acquisition device group is used to capture images of the front or back of the fabric. The system adopts the following technical solution:

[0047] The image information receiving module is used to receive image information sent by the image acquisition device group. The image information includes several actual images with cloth areas and the identification code of the image acquisition device corresponding to the actual image of the target.

[0048] The standard image retrieval module is used to retrieve the standard image corresponding to the target's identity code from a preset fabric image reference database;

[0049] The similarity comparison module is used to compare the actual image with the standard image to obtain the actual similarity value;

[0050] The fabric defect judgment module is used to determine whether there are defects in the fabric area of ​​the actual image if the actual similarity value is less than the preset similarity standard value.

[0051] The fabric defect warning module is used to send warning information related to defects in the fabric area shown in the actual image to the user terminal.

[0052] By adopting the above technical solution, the automatic detection of fabric defects during the fabric production process replaces manual inspection, saving time and effort and increasing detection efficiency. After detecting defects, an early warning message is sent to the user, allowing the user to be aware of the defects in a timely manner. The user can then repair the double-needle bed warp knitting machine according to the defects, minimizing the continuous production of defective fabrics and helping to reduce the amount of defective fabrics.

[0053] Thirdly, this application provides a computer device that adopts the following technical solution: it includes a memory and a processor, wherein the memory stores a computer program that can be loaded by the processor and executed as described above for any of the fabric defect detection methods.

[0054] Fourthly, this application provides a computer-readable storage medium, which employs the following technical solution: storing a computer program capable of being loaded by a processor and executing any of the above-mentioned fabric defect detection methods.

[0055] In summary, this application includes at least one of the following beneficial technical effects:

[0056] 1. Through automatic detection of fabric defects, the detection efficiency is high. After detecting the presence of defects, an early warning message is sent to the user terminal so that the user can be informed of the defects in a timely manner. Then, the user can repair the double needle bed warp knitting machine according to the defects, so as to avoid the double needle bed warp knitting machine from continuously producing fabrics with defects, which helps to reduce the amount of defective fabrics.

[0057] 2. By taking pictures of the front or back of the fabric with several image acquisition devices, it is possible to avoid blind spots in the shooting and thus avoid missed detection, resulting in better defect detection.

[0058] 3. By controlling the backlighting devices for supplementing the light source on the front of the fabric and the backlighting devices for supplementing the light source on the back of the fabric to work alternately, it helps to avoid overexposure phenomena in the fabric images captured and processed by the image acquisition devices on the front and back sides, thus improving the accuracy of fabric defect detection. Attached Figure Description

[0059] Figure 1 This is a flowchart of the fabric defect detection method in the embodiments of this application.

[0060] Figure 2 This is a structural block diagram of the fabric defect detection system in the embodiments of this application.

[0061] Explanation of reference numerals in the attached diagram: 201, Image information receiving module; 202, Standard image retrieval module; 203, Similarity comparison module; 204, Fabric defect judgment module; 205, Fabric defect early warning module. Detailed Implementation

[0062] The following is in conjunction with the appendix Figure 1-2 This application will be described in further detail.

[0063] This application discloses a method for detecting fabric defects. The method is applied to a double-needle bed warp knitting machine. The double-needle bed warp knitting machine is equipped with an image acquisition device group, which includes several front image acquisition devices and several back image acquisition devices arranged along the width direction of the fabric. Each front and back image acquisition device has a corresponding identification code. The front image acquisition devices are used to capture images of the front side of the fabric, and the back image acquisition devices are used to capture images of the back side of the fabric. The double-needle bed warp knitting machine is equipped with a controller, which contains a fabric defect detection system. This system controls the operation of the front and back image acquisition devices and identifies whether there are defects on the fabric surface based on the captured images.

[0064] like Figure 1 As shown, the method includes the following steps:

[0065] S101 receives image information sent by the image acquisition device group.

[0066] Specifically, the system sends shooting commands to several front-facing image acquisition devices or several back-facing image acquisition devices in the image acquisition device group. Upon receiving the commands, these devices photograph the fabric and transmit the captured images back to the system. The image information includes several actual images showing the fabric area and the identification code of the image acquisition device corresponding to each actual image. Based on the identification code, the system can identify the corresponding image acquisition device and then determine the specific location of the fabric within the captured image.

[0067] S102, retrieve the standard image corresponding to the target identity code from the preset fabric image reference database.

[0068] Specifically, the system pre-establishes a fabric image reference database. This database stores identification codes corresponding to several image acquisition devices, along with corresponding standard images. The standard images are taken by the image acquisition device corresponding to the target identification code, and the fabric area in these images is free of defects. After receiving image information, the system can retrieve the standard image corresponding to any identification code from the pre-set fabric image reference database.

[0069] S103, compare the actual image with the standard image to obtain the actual similarity value.

[0070] Specifically, after the system retrieves the standard image corresponding to the target identity code, it compares the actual image with the standard image to obtain the actual similarity value between the actual image and the standard image.

[0071] S104. If the actual similarity value is less than the preset similarity standard value, it is determined that there are defects in the fabric area of ​​the actual image.

[0072] Specifically, the system pre-stores similarity standard values, which correspond to the similarity between the actual image and the standard image when they are similar. After retrieving the similarity standard values, the system compares the actual similarity value with the preset similarity standard value. If the actual similarity value is less than the preset similarity standard value, it indicates that there are defects in the fabric area of ​​the actual image.

[0073] S105, send a warning message to the user terminal related to the presence of defects in the fabric area in the actual image.

[0074] Specifically, the user terminal can be a mobile terminal such as a mobile phone, computer, or iPad. In this embodiment, the user terminal can be the mobile terminal of the operator and the mobile terminal of the equipment back-end management personnel. After the system determines that there is a defect in the fabric area in the actual image, it sends a warning message to the mobile terminals of the operator and the back-end management personnel. For example, the warning message could be: "Hello, there is a defect in the fabric area in the actual image taken by the image acquisition device corresponding to the target identity code. Please repair the equipment immediately!"

[0075] In one embodiment, considering that the large shooting range of the image acquisition device results in numerous device structures in the captured images, thus affecting the accuracy of the similarity comparison results between the actual images and the standard images, the following processing can be performed before comparing the actual images with the standard images:

[0076] The system pre-stores image cropping rules. For example, the image cropping rule can be set to crop the image at the center of the actual image, and the length of the cropped image is the same as the length of the actual image, while the width of the cropped image is 1 / 5 of the width of the actual image.

[0077] The system then crops the actual image according to preset image cropping rules to obtain a corrected image, which is located at the center of the actual image. The corrected image is then set as the actual image. Subsequently, the image compared with the standard image by the system is actually the corrected image. Because the size of the corrected image is small, it helps to reduce the probability of other debris in the actual image compared with the standard image, which helps to improve the accuracy of the similarity comparison results between the final actual image and the standard image, minimizes the probability of misjudgment, and improves the accuracy of defect detection results.

[0078] In one embodiment, considering the possibility of obstacles obstructing the lens of the image acquisition device, resulting in a low similarity between the actual image and the standard image, the handling of determining that the fabric area in the actual image has defects can be as follows:

[0079] The system pre-stores rules for extracting the blemish outline range. In this embodiment, the system can extract the actual blemish outline region corresponding to the difference between the actual image and the standard image in the actual image according to the preset rules for extracting the blemish outline range. Then, the system identifies the actual blemish features of the actual blemish outline region. The actual blemish features specifically include the actual blemish color features and the actual blemish texture features.

[0080] In this embodiment, a defect category database is pre-established in the system. This database stores all actual defect characteristics that may exist on the fabric surface during the fabric production process, such as broken threads or multiple threads. When broken threads or multiple threads are present on the fabric surface, the color and texture of the defective area differ from those of the normal fabric surface.

[0081] The system then queries a pre-defined defect category database to check for actual defect features. If a defect is found, the system determines that the fabric area in the actual image contains a defect. If no defect is found, it indicates that the fabric area in the actual image may contain an obstacle. The system pre-stores an obstacle feature database, which contains feature information corresponding to the structure of a double-needle bed warp knitting machine that may obstruct the image acquisition device's lens. For example, the connecting post on the double-needle bed warp knitting machine that obstructs the image acquisition device's lens is black and has a smooth texture.

[0082] The system then searches the preset obstacle feature database for actual defect features. If so, it indicates that there is an obstacle in the fabric area of ​​the actual image. If not, it indicates that the lens surface of the image acquisition device is dirty, which affects the accuracy of the similarity comparison results between the actual image and the standard image.

[0083] In one embodiment, considering that obstacles in the fabric area of ​​the captured actual image may affect the accuracy of the similarity comparison result between the actual image and the standard image, after determining that there are obstacles in the fabric area of ​​the actual image, the following steps can also be implemented:

[0084] If the system determines that there is an obstacle in the fabric area of ​​the actual image, the system sends a reminder message related to the obstacle to the user terminal. In this embodiment, the reminder message can be a text message. For example, the text message can be pre-edited text content, such as "Hello, there is an obstacle in the actual image captured by the image acquisition device corresponding to the target identity code. Please confirm whether there is an obstacle in the actual image." The user can then confirm the obstacle in the actual image based on the text message reminder, and the user can also use the mobile terminal to provide feedback to the system on whether the obstacle in the actual image is part of the equipment structure of a double needle bed warp knitting machine.

[0085] After the system sends an SMS reminder to the user, within a preset first time period, say 1 minute, if the system receives confirmation from the user that the obstacle in the actual image is a double-needle bed warp knitting machine, it means that there is no defect in the fabric area of ​​the actual image. Rather, the obstacle is obstructed by the structure of the equipment in front of the lens of the image acquisition device, which affects the accuracy of the similarity comparison results between the actual image and the standard image.

[0086] Subsequently, to improve the accuracy of similarity comparison results, the actual image is updated to a standard image by modifying the fabric area in the actual image. The standard image is an image where the fabric area has obstacles and the fabric area in the standard image is free of defects. This improves the accuracy of subsequent similarity comparisons between the actual image and the standard image. Deep learning of the standard images in the fabric image reference database is implemented, which helps to improve the accuracy of defect detection results.

[0087] In one embodiment, considering the impact of fabric defects on fabric production quality, the steps for sending warning information related to the presence of defects in the fabric area shown in the actual image to the user can be as follows:

[0088] The system has a pre-established defect type database, which stores several actual defect features and their corresponding defect names. After determining that a defect exists in the fabric area of ​​an actual image, the system retrieves the defect name corresponding to the actual defect feature from the pre-established defect type database.

[0089] The system also has a pre-established database of allowed early warning defects. This database stores the names of defects that affect the quality of fabric production. If a defect name is not found in the pre-established database, it means that the defects in the fabric area in the actual picture will not affect the quality of the fabric production. Therefore, the system will not perform any further processing so that the warp knitting machine can continue to operate.

[0090] If the system finds the defect name in the preset allowed early warning defect database, it indicates that the defect features in the fabric area of ​​the actual image will affect the quality of the produced fabric. The system then controls the double-bed warp knitting machine to stop operating, ensuring timely shutdown. Simultaneously, it sends an early warning message related to the defect in the fabric area of ​​the actual image to the user, allowing the user to be promptly informed of the defect. This approach ensures that the system automatically stops the warp knitting machine upon detecting a defect, avoiding the need for the user to manually stop the machine after being notified. This helps prevent the warp knitting machine from continuously operating and producing more defective products, thus reducing the amount of defective fabric.

[0091] In this embodiment, the system also pre-stores an equipment fault alarm database. The equipment fault alarm database pre-stores all defect names located in the allowed warning defect database and the possible equipment faults of the warp knitting machine corresponding to the target defect name. When the system sends warning information related to the presence of defects in the fabric area to the user terminal, the warning information may also include the equipment fault corresponding to the defect name, so that the operator can quickly find the equipment fault and carry out equipment maintenance based on the warning information, which helps to improve maintenance efficiency and avoid affecting the production efficiency of the fabric.

[0092] In one embodiment, considering the possibility that dust adhering to the lens of the image acquisition device may result in a low similarity between the captured image and the standard image, the double-needle bed warp knitting machine is equipped with several cleaning air nozzles, each corresponding to one of the image acquisition devices. An external air source is connected to the cleaning air nozzles, and the air outlets of the cleaning air nozzles face the lens surface of the image acquisition device. The steps for determining whether the lens surface of the image acquisition device is dirty can be as follows:

[0093] If the system does not find any actual defect features in the preset obstacle feature database, the system first identifies the clarity of the actual image to obtain the actual clarity. The system pre-stores a clarity standard range, which corresponds to the clarity range of images taken when the lens of the image acquisition device is not dirty. Then the system determines whether the actual clarity is within the preset clarity standard range. If not, the system determines that the surface of the lens of the image acquisition device is dirty.

[0094] The system then controls the cleaning nozzle to supply air to the lens surface of the image acquisition device corresponding to the target image, allowing the cleaning nozzle to blow air onto the lens surface to remove dust. Then, after a preset second time period—for example, 10 seconds after the system controls the cleaning nozzle to blow away dust from the lens surface—the system sends a fabric image capture command to the image acquisition device corresponding to the identification code. This causes the target image acquisition device to capture the fabric again, and then transmit the updated image of the fabric back to the system.

[0095] After receiving the updated photo, the system re-evaluates the updated image's sharpness. If the updated sharpness falls within a preset sharpness standard range, it indicates that the lens surface of the image acquisition device is dirty, resulting in low similarity between the actual image and the standard image. If the updated sharpness falls outside the preset sharpness standard range, it indicates that the lens surface of the image acquisition device is not dirty, but the lens may be damaged.

[0096] The above method, which cleans the lens surface to verify whether the blurriness is due to dust, is accurate when the actual image is detected to be unclear. Furthermore, cleaning the dust from the lens surface helps prevent it from affecting the clarity of subsequent images of the fabric captured by the image acquisition device. This also helps prevent dirt on the lens surface of the image acquisition device from causing low similarity between the actual image and the standard image, thus improving the accuracy of verifying whether there are defects in the fabric area of ​​the actual image.

[0097] In one embodiment, considering that the actual image captured by the image acquisition device is unclear due to dim lighting when photographing the fabric, the double needle bed warp knitting machine is equipped with several backlight devices. The backlight device group is used to supplement the light source for the front or back of the fabric. Before receiving the image information sent by the image acquisition device group, the following steps can also be implemented:

[0098] When the system detects the operation of the double needle bed warp knitting machine, it first controls the backlight device that provides supplementary light to the reverse side of the fabric to start supplementary lighting, and controls the backlight device that provides supplementary light to the front side of the fabric to stop supplementary lighting. At the same time, the system sends front shooting instructions to several front image acquisition devices. The front shooting instructions include the front identification code corresponding to the target front image acquisition device. Then, the front image acquisition device corresponding to the target front identification code receives the shooting instructions and performs the shooting operation on the front side of the fabric.

[0099] Then the system controls the backlight device that provides supplemental light to the reverse side of the fabric to stop supplemental lighting and controls the backlight device that provides supplemental light to the front side of the fabric to start supplemental lighting. At the same time, the system sends reverse shooting instructions to several reverse image acquisition devices. The reverse shooting instructions include the reverse identification code corresponding to the target reverse image acquisition device. Then, after receiving the shooting instructions, the reverse image acquisition device corresponding to the reverse identification code performs shooting operations on the reverse side of the fabric.

[0100] Using the above shooting method, whether the front image acquisition device or the back image acquisition device is shooting the fabric, the corresponding backlight device is controlled to supplement the light on the fabric surface. This helps to avoid the fabric surface images being unclear due to dim lighting in the working environment. Furthermore, during the front image acquisition device's shooting of the fabric surface, the backlight device supplements the light on the back of the fabric surface. Conversely, when the back image acquisition device is shooting the fabric surface from the back, the backlight device supplements the light on the front of the fabric surface. This helps to illuminate the texture of the fabric surface, making the texture of the fabric area in the captured image clear. This helps to improve the accuracy of judging whether there are defects on the fabric surface, and at the same time, it helps to avoid overexposure caused by supplementing the light on the front of the fabric surface when the front image acquisition device is shooting the fabric surface.

[0101] Ultimately, when taking photos of both sides of the fabric, the backlighting devices that supplement the light source on the front of the fabric and those that supplement the light source on the back of the fabric work alternately. This helps to avoid overexposure in the processed fabric images and prevents the images from being blurry due to dim lighting, which would affect the accuracy of the similarity comparison between the actual image and the standard image, and further improve the accuracy of fabric defect detection.

[0102] The implementation principle of this application embodiment is as follows: During operation, the double-needle bed warp knitting machine first uses an image acquisition device group to capture real-time images of the front or back of the fabric. Then, the system retrieves the standard image corresponding to the identification code of the image acquisition device corresponding to the actual image from a preset fabric image reference database. If the actual similarity value obtained by comparing the actual image and the standard image is less than the preset similarity standard value, it indicates that there is a defect in the fabric area of ​​the actual image. This realizes automatic detection of fabric defects, replacing manual inspection, saving time and effort. Then, the system sends warning information related to the defect in the fabric area of ​​the actual image to the user terminal, so that the user can be informed of the defect in the fabric in a timely manner. The user can then repair the double-needle bed warp knitting machine according to the defect, minimizing the production of defective fabrics and helping to reduce the amount of defective fabrics.

[0103] Based on the above method, this application also discloses a fabric defect detection system. This system is applied to a double-needle bed warp knitting machine, which is equipped with an image acquisition device group. The image acquisition device group is used to capture images of the front or back of the fabric. The system includes:

[0104] The image information receiving module 201 is used to receive image information sent by the image acquisition device group. The image information includes several actual pictures with cloth areas and the identification code of the image acquisition device corresponding to the actual picture of the target.

[0105] The standard image retrieval module 202 is used to retrieve the standard image corresponding to the identity code from the preset fabric image reference database;

[0106] The similarity comparison module 203 is used to compare the actual image with the standard image to obtain the actual similarity value;

[0107] The fabric defect judgment module 204 is used to determine whether there are defects in the fabric area of ​​the actual image if the actual similarity value is less than the preset similarity standard value.

[0108] The fabric defect warning module 205 is used to send warning information related to defects in the fabric area in the actual image to the user terminal.

[0109] In one embodiment, the similarity comparison module 203 is further configured to:

[0110] According to the preset image cropping rules, the actual image is cropped to obtain the corrected image; the corrected image is then set as the actual image.

[0111] In one embodiment, the fabric defect detection module 204 is further configured to:

[0112] Based on preset blemish contour range extraction rules, the actual blemish contour regions corresponding to the differences between the actual image and the standard image are extracted from the actual image. The actual blemish features of the actual blemish contour regions are identified, including actual blemish color features and actual blemish texture features. A preset blemish category database is searched to see if the actual blemish features exist. If yes, it is determined that the fabric area in the actual image has a blemish. If not, a preset obstacle feature database is retrieved. A preset obstacle feature database is searched to see if the actual blemish features exist. If yes, it is determined that the fabric area in the actual image has an obstacle. If not, it is determined that the lens surface of the image acquisition device is dirty.

[0113] In one embodiment, the fabric defect detection module 204 is further configured to:

[0114] Send a reminder message to the user terminal regarding the presence of an obstacle in the fabric area shown in the actual image; if, within a preset first time period, confirmation information is received from the user terminal that the obstacle in the actual image is the equipment structure of a double-needle bed warp knitting machine, then the actual image is updated to a standard image.

[0115] In one embodiment, the fabric defect early warning module 205 is further used for:

[0116] Retrieve the defect name corresponding to the actual defect characteristics from the preset defect type database; check if the defect name exists in the preset allowed warning defect database; if so, control the double needle bed warp knitting machine to stop operation and send warning information related to the presence of defects in the fabric area in the actual picture to the user terminal.

[0117] In one embodiment, the fabric defect detection module 204 is further configured to:

[0118] The system identifies the actual sharpness of the actual image; determines whether the actual sharpness is within a preset sharpness standard range; if not, it determines that the lens surface of the image acquisition device is dirty; it replenishes air to the lens surface of the target image acquisition device; after a preset second time period, it sends a cloth shooting command to the image acquisition device corresponding to the identity code; it obtains the updated image returned by the image acquisition device corresponding to the identity code; it identifies the updated sharpness of the updated image; if the updated sharpness is within a preset sharpness standard range, it determines that the lens surface of the image acquisition device is dirty.

[0119] In one embodiment, the image information receiving module 201 is further configured to:

[0120] The system controls the backlighting device that provides supplemental light to the reverse side of the fabric to start supplemental lighting and controls the backlighting device that provides supplemental light to the front side of the fabric to stop supplemental lighting; it sends frontal shooting instructions to several frontal image acquisition devices, the frontal shooting instructions including the frontal identification code corresponding to the target frontal image acquisition device; it controls the backlighting device that provides supplemental light to the reverse side of the fabric to stop supplemental lighting and controls the backlighting device that provides supplemental light to the front side of the fabric to start supplemental lighting; it sends backal shooting instructions to several backal image acquisition devices, the backal shooting instructions including the backal identification code corresponding to the target backal image acquisition device.

[0121] This application also discloses a computer device.

[0122] Specifically, the device includes a memory and a processor, the memory storing a computer program that can be loaded by the processor and executed as described above for detecting fabric defects.

[0123] This application also discloses a computer-readable storage medium.

[0124] Specifically, the computer-readable storage medium stores a computer program that can be loaded by a processor and executed as described above for detecting fabric defects. The computer-readable storage medium includes, for example, various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0125] The embodiments described in this specific implementation are preferred embodiments of this application and are not intended to limit the scope of protection of this application. Therefore, all equivalent changes made in accordance with the structure, shape and principle of this application should be covered within the scope of protection of this application.

Claims

1. A method for detecting fabric surface defects, characterized in that, This method is applied to a double-needle bed warp knitting machine, which is equipped with an image acquisition device group. The image acquisition device group is used to capture images of the front or back of the fabric. The method includes: Receive image information sent by the image acquisition device group, the image information including several actual images with cloth areas and the identification code of the image acquisition device corresponding to the target actual image; Retrieve the standard image corresponding to the target's identity code from the preset fabric image reference database; The actual image is compared with the standard image to obtain the actual similarity value; If the actual similarity value is less than the preset similarity standard value, then it is determined that there are defects in the fabric area of ​​the actual image, including: Based on the preset flaw outline range extraction rules, extract the actual flaw outline region corresponding to the difference between the actual image and the standard image in the actual image; Identify the actual defect features of the actual defect outline region, the actual defect features including actual defect color features and actual defect texture features; The system queries a preset defect category database to determine if the actual defect feature exists. If it does, it determines that the fabric area in the actual image has a defect. If not, it retrieves a preset obstacle feature database. The system queries a preset obstacle feature database to determine if the actual defect feature exists. If it does, it determines that there is an obstacle in the fabric area of ​​the actual image. If not, it determines that there is dirt on the lens surface of the image acquisition device. Send a warning message to the user terminal regarding the presence of defects in the fabric area shown in the actual image.

2. The fabric defect detection method according to claim 1, characterized in that, Before comparing the actual image with the standard image, the method further includes: According to the preset image cropping rules, the actual image is cropped to obtain the corrected image; Set the corrected image as the actual image.

3. The fabric defect detection method according to claim 1, characterized in that, After determining that there is an obstacle in the fabric area of ​​the actual image, the method further includes: Send a notification message to the user's device that indicates an obstacle exists in the fabric area shown in the actual image; If, within a preset first time period, a confirmation message is received from the user terminal that the obstacle in the actual image is the equipment structure of a double-needle bed warp knitting machine, then the actual image will be updated to a standard image.

4. The fabric defect detection method according to claim 1, characterized in that, Sending warning information to the user terminal related to defects in the fabric area shown in the actual image includes: Retrieve the defect name corresponding to the actual defect characteristics from the preset defect type database; If the defect name is found in the preset database of allowed early warning defects, the double needle bed warp knitting machine is stopped and an early warning message related to the defect in the fabric area in the actual picture is sent to the user terminal.

5. The fabric defect detection method according to claim 1, characterized in that, The determination that the lens surface of the image acquisition device is dirty includes: Identify the actual sharpness of the actual image; Determine whether the actual sharpness is within the preset sharpness standard range; if not, determine that the lens surface of the image acquisition device is dirty. Supplement the air source to the lens surface of the image acquisition device described in the target; After a preset second time period, a cloth shooting instruction is sent to the image acquisition device corresponding to the identity code; Obtain the updated image returned by the image acquisition device corresponding to the identity code; Identify the updated sharpness of the updated image; If the updated sharpness is within the preset sharpness standard range, it is determined that the lens surface of the image acquisition device is dirty.

6. The fabric defect detection method according to claim 1, characterized in that, The image acquisition device group includes several front image acquisition devices for capturing images of the front of the fabric and several back image acquisition devices for capturing images of the back of the fabric. The double-needle bed warp knitting machine is equipped with several backlight devices, which are used to supplement the light source for the front or back of the fabric. Before receiving the image information sent by the image acquisition device group, the following is also included: The backlight device that controls the supplementary light source on the reverse side of the fabric to start the supplementary lighting operation, and the backlight device that controls the supplementary light source on the front side of the fabric to stop the supplementary lighting operation; Send a frontal shooting command to a plurality of the aforementioned frontal image acquisition devices, wherein the frontal shooting command includes a frontal identification code corresponding to the target frontal image acquisition device; The backlight device that provides supplementary light source to the reverse side of the fabric stops its supplementary lighting operation, and the backlight device that provides supplementary light source to the front side of the fabric starts its supplementary lighting operation. Send a reverse image capture command to a plurality of the reverse image acquisition devices, wherein the reverse image capture command includes a reverse identity code corresponding to the target reverse image acquisition device.

7. A fabric defect detection system, characterized in that, This system is applied to a double-needle bed warp knitting machine, which is equipped with an image acquisition device group. The image acquisition device group is used to capture images of the front or back of the fabric. The system includes: The image information receiving module (201) is used to receive image information sent by the image acquisition device group. The image information includes several actual pictures with cloth areas and the identity code of the image acquisition device corresponding to the actual picture of the target. The standard image retrieval module (202) is used to retrieve the standard image corresponding to the target's identity code from a preset fabric image reference database; The similarity comparison module (203) is used to compare the actual image with the standard image to obtain the actual similarity value; The fabric defect judgment module (204) is used to determine whether there are defects in the fabric area of ​​the actual image if the actual similarity value is less than the preset similarity standard value; it is also used to extract the actual defect contour area corresponding to the difference between the actual image and the standard image in the actual image according to the preset defect contour range extraction rules; identify the actual defect features of the actual defect contour area, the actual defect features include actual defect color features and actual defect texture features; query whether there are actual defect features in the preset defect category database; if yes, determine whether there are defects in the fabric area of ​​the actual image; if no, retrieve the preset obstacle feature database; query whether there are actual defect features in the preset obstacle feature database; if yes, determine whether there are obstacles in the fabric area of ​​the actual image; if no, determine whether there is dirt on the lens surface of the image acquisition device. The fabric defect warning module (205) is used to send warning information related to the presence of defects in the fabric area in the actual image to the user terminal.

8. A computer device, characterized in that: It includes a memory and a processor, wherein the memory stores a computer program that can be loaded by the processor and executed as any one of the fabric defect detection methods as claimed in claims 1 to 6.

9. A computer-readable storage medium, characterized in that, The computer program is stored and can be loaded by a processor and executed as any one of the fabric defect detection methods as claimed in claims 1 to 6.

Images