Visual intelligent detection device and method for tensile strain of fabric

Abstract

The invention discloses a fabric tensile strain visual intelligent detection method. The method comprises the following steps: (1) acquiring a fabric tensile video; (2) video sequence frame preprocessing: performing batch processing on the acquired fabric stretching video, performing image extraction according to a time sequence, and preprocessing a video sequence frame image; (3) deformation characteristic index extraction and calculation, including real-time characteristic point extraction and matching and deformation calculation, to obtain elongation, narrowest shrinkage and Poisson's ratio; (4) improving an image correlation analysis algorithm; according to the innovative detection method provided by the invention, an advanced machine vision technology and image correlation analysis are fused, and a fabric performance visual detection module is improved. Deformation of the fabric in the stress stretching process is monitored in real time, microscopic strain characteristics are accurately captured, and comprehensive evaluation of the fabric performance is achieved. By automatically calculating the morphological characteristic index and Poisson's ratio of the fabric and visualizing the in-plane strain distribution, the detection efficiency and precision are greatly improved.

Description

Technical Field

[0001] The invention belongs to the field of fabric mechanical property detection, and specifically relates to a visual intelligent detection device and method for fabric tensile strain. Background Art

[0002] The application of textile materials has gradually expanded to the fields of smart clothing and high-performance fabrics, which has put forward higher requirements for the analysis of basic mechanical properties of fabrics, prompting new evaluation methods to attract much attention. The tensile properties of fabrics are an important part of basic mechanical properties and are the primary evaluation indicators for evaluating and developing new products. The main test indicators include tensile strength, elongation at break, elastic modulus, yield strength, elastic recovery rate, tensile stress-strain curve, and rupture elongation. Among them, tensile strain, as an important indicator, helps to evaluate the elasticity and deformation capacity of fabrics, and is used in combination with breaking strength and elongation at break to fully understand the mechanical properties of fabrics. The contact measurement methods of material mechanical properties mainly include electrical measurement and tensile measurement. In the electrical measurement method, the contact point between the extensometer arm and the sample is prone to excessive stress concentration, which is greatly affected by contact, and the detection efficiency and stability are poor; although the fabric tensile measuring instrument can measure the tensile force as a whole, it is difficult to obtain the detailed strain characteristics in the surface, and it is impossible to fully quantify the complex deformation and local stress distribution of the fabric. Therefore, it is urgent to develop a more intuitive, simple, and cross-scale test method.

[0003] In the field of textile materials, the application of machine vision technology and digital image correlation technology is gradually increasing. However, these technologies are still insufficient in visually representing the tensile mechanical properties of different fabrics and treatment methods. For example, in terms of machine vision-based detection systems, the video extensometer and image analysis technology used by Marian Klasztorny et al. and Rodrigues A et al. provide valuable data in strain and stress detection, but are affected by equipment limitations and friction factors. In terms of digital image correlation technology, the measurement method developed by Wang Mingliang et al. can obtain microscopic deformation information of yarns and fabrics, and López-Santos et al. demonstrated the advantages of DIC technology in composite material strain measurement. Although these technologies show potential, they still need to be further optimized and verified to be better applied to fabric strain detection.

[0004] The currently published related patents focus on the innovation of detection devices, including: 1) Patent with publication number CN109765099A discloses a method for measuring the tensile properties of fabrics, improves the fabric clamping device, and reduces the unevenness of the fracture surface; 2) Patent with publication number CN210533865U discloses a clamp and test system for fabric stretching, and a device for improving the stability of fabric clamping; 3) Patent with publication number CN117054233A discloses a warp knitted article tensile performance detection equipment, which enhances the symmetry and firmness of the warp knitted fabric clamping and improves the multi-directional detection capability.

[0005] The focus is on the innovation of inspection methods: 1) Patent No. CN102507324B discloses a variable speed tensile performance tester for fabrics, which simulates the actual use state of the fabric and draws a load-elongation curve; 2) Patent No. CN217277410U discloses a three-dimensional tensile performance test device for fabric composite materials, providing a digital representation of the stretching effect; 3) Patent No. CN117825152A discloses a material local strain measurement device and method based on image processing, which combines image processing technology with biaxial tensile test to quickly and accurately measure the elastic modulus, Poisson's ratio and shear modulus of flexible fabric materials; 4) Patent No. CN118624367A discloses a method for controlling the laser irradiation point to track the marking line for non-contact measurement of elongation at break, which effectively solves the compatibility problem between the extensometer and various tensile machine software and data interfaces, and is suitable for a variety of inspection scenarios such as small deformation materials and large deformation materials.

[0006] In summary, the existing technology has several deficiencies in the detection of tensile strain of textile materials. Current technology usually relies on manual operation and surface analysis methods, which is difficult to fully reflect the overall performance of the material. This limitation leads to insufficient in-depth and accurate evaluation of the material. Secondly, the existing testing devices have problems with measurement accuracy. For example, the instability of the clamping and fixing effects may lead to unreliable data. In addition, the operating procedures and data processing of the existing detection methods are relatively complicated, which not only affects the test efficiency, but also limits the real-time performance. Furthermore, the existing technology is often unable to fully obtain detailed local or in-plane features when evaluating the performance of textile materials, and it is difficult to accurately reflect the internal stress conduction and distribution caused by the structural characteristics of the fabric. Especially in the process of fabric stretching, the research on visual representation methods is still insufficient. Therefore, refined representation and measurement are particularly urgent and important for the research and development and application of fabric materials. Summary of the invention

[0007] The purpose of the present invention is to provide a visual intelligent detection device and method for fabric tensile strain. Aiming at the defects in the prior art, an innovative detection method is proposed, which integrates advanced machine vision technology and image correlation analysis to improve the visual detection module of fabric performance. The deformation of the fabric during the stress stretching process is monitored in real time, the microscopic strain characteristics are accurately captured, and a comprehensive evaluation of the fabric performance is achieved. By automatically calculating the morphological characteristic index and Poisson's ratio of the fabric and visualizing the in-plane strain distribution, the detection efficiency and accuracy are greatly improved.

[0008] In order to solve the above technical problems, the following technical solutions are adopted:

[0009] A visual intelligent detection method for fabric tensile strain, characterized by comprising the following steps:

[0010] (1) Fabric stretching video acquisition;

[0011] (2) Video sequence frame preprocessing: The collected fabric stretching videos are batch processed, images are extracted according to the time series, and the video sequence frame images are preprocessed;

[0012] (3) Extraction and calculation of deformation feature indicators: including real-time feature point extraction and matching and deformation calculation to obtain elongation, contraction at the narrowest point and Poisson's ratio;

[0013] (4) Improvement of image correlation analysis algorithm: In the tensile test, the improved image correlation analysis algorithm is applied, and the initial frame image is used as the reference image and other images are used as deformation images to analyze the in-plane dynamic deformation of the fabric during the stretching process. A deep learning model is introduced for feature matching to improve the accuracy and speed of deformation detection. At the same time, a performance rating mechanism is added to objectively evaluate the mechanical properties of different fabrics, and the deformation process is dynamically displayed through data visualization technology.

[0014] After optimization, the fabric samples need to be pre-treated, including:

[0015] 1) Sample cutting: Cut the fabric into standard size according to relevant standards or user-specified size.

[0016] 2) Speckle production: Based on the sample specifications, a speckle pattern that meets the requirements of high contrast, randomness, non-directionality, and stability is automatically generated through computer simulation; an acrylic plate is used to perform blank processing based on the speckle position, and then a roller brush is used to dip ink to transfer the speckle pattern to the surface of the fabric sample to complete the speckle production;

[0017] 3) Temperature adjustment and debugging treatment: The fabric sample after the above treatment is placed in a constant temperature and humidity environment for 24 hours to adjust the temperature and humidity to stabilize the physical properties of the fabric sample.

[0018] After optimization, the step (1) uses a visual detection device for fabric tensile strain to collect fabric stretching video, the device includes a base, an adjustable camera bracket, a movable track, an upper clamping device, a lower clamping device and a computer, the adjustable camera bracket is used to install the camera, and is detachable, foldable, and adjustable in height and angle; the lower clamping device and the upper clamping device respectively clamp the two ends of the fabric for stretching the fabric; the movable track is connected with a stretching drive motor, a movable device and a drive switch, the stretching drive motor drives the movable device to move on the movable track, and the movable device is connected to the upper clamping device to drive the upper clamping device to move on the movable track; the computer is used to input and output experimental data and visual capture images in real time;

[0019] The collection process includes: adjusting the distance between the upper clamping device and the lower clamping device to clamp the fabric sample between the two; adjusting the height of the device and the distance between the camera and the fabric sample to ensure that the fabric deformation image during the entire stretching process is completely recorded.

[0020] Preferably, a storage slot is provided at the position where the adjustable camera holder is installed on the base, and the storage slot is used to store the adjustable camera holder; a groove is also provided on the base, and the groove is used to place a background board; a light strip is also installed on the movable track, and the light strip is used to provide a light source.

[0021] After optimization, the step (2), video sequence frame image preprocessing includes:

[0022] 1) Region of interest extraction: Through binarization processing and maximum bounding rectangle algorithm, the region of interest where the fabric is located in each set of video frames is preliminarily determined to reduce computational redundancy.

[0023] 2) Image enhancement: Convert the cropped region of interest to a grayscale image and apply adaptive histogram equalization to enhance the contrast;

[0024] 3) Contour detection and mask creation: Binarize the enhanced image, identify the largest outer contour, and create a minimum bounding rectangle mask based on it.

[0025] After optimization, the step (3) further includes pixel calibration: converting the pixel value into the actual physical size in advance.

[0026] After optimization, the step (3) is real-time feature point extraction and matching: extracting key feature points from the first frame image that has not been deformed, and using an improved image correlation analysis algorithm to compare the deformed image with the reference image to obtain the displacement field; at the same time, during the deformation image analysis process, using the trained model to perform real-time feature point extraction and matching to identify and track the feature points on the fabric surface.

[0027] After optimization, the step (3), deformation calculation: calculate the deformation of each feature point, including the elongation, the contraction at the narrowest point, and the Poisson's ratio:

[0028] 1) Elongation calculation:

[0029] ΔL i =L i -L0

[0030] Where: L0 is the initial length of the effective area; L i is the length of the i-th interval frame image after stretching;

[0031] 2) Calculation of shrinkage at the narrowest point:

[0032] ΔW i =W i -W0

[0033] Where: W0 is the initial width, W i is the width after stretching;

[0034] 3) Poisson's ratio calculation:

[0035]

[0036] After optimization, the step (4) uses an integrated data acquisition and processing module to integrate the step (2) and the step (3) into the module of the step (4), and performs the following operations:

[0037] 1) Image registration: processing the acquired frame images according to step (3) and step (4), taking the first frame image as a reference image and the remaining frame images as deformation images, so as to analyze the in-plane deformation of the fabric during the stretching process;

[0038] 2) Correlation calculation: The zero-mean normalized least square distance correlation function ZNSSD is used as the correlation criterion, and its calculation method is:

[0039]

[0040] Where: f(x i ,y i ), p(x′ i ,y′ i ) are the grayscale values ​​of the pixels in the reference sub-region and the current sub-region respectively; f m 、p m are the average gray values ​​corresponding to the reference sub-area and the deformed sub-area respectively; x i ,y iare the horizontal and vertical coordinates of the pixels in the corresponding deformed sub-area in the whole image; i and j are the numbers of the pixels in the sub-area; σ1 and σ2 are the grayscale standard deviations of the reference sub-area and the current sub-area, indicating the degree of change or the breadth of the distribution of the grayscale value of the image; M and N are the height and width of the image, respectively; MN is the number of pixels in the image; f and p are the reference sub-area and the current sub-area, respectively, with a size of M×N; It means to sum the positions of all pixels in the image;

[0041] 3) Iterative optimization and model training: Use the Newton-Raphson iteration method to optimize the calculation, and use the labeled data set to train the deep learning model;

[0042] 4) Strain analysis and performance rating: Real-time calculation of the strain field of the fabric, generation of strain distribution diagram, and mechanical performance rating based on stress-strain curve and Poisson’s ratio;

[0043] 5) Data visualization technology: Using data visualization technology, the deformation process is dynamically displayed, animation effects are generated, the deformation state of the fabric during the stretching process is clearly presented, and a visualization report is generated, including stress-strain curves and Poisson's ratio;

[0044] 6) System optimization and adaptability: Adaptive learning function adjusts processing algorithm parameters based on real-time data feedback.

[0045] After optimization, the specific steps of step 1) image registration include:

[0046] a. Select the first frame image at the beginning of stretching as the reference image, and the last frame image as the image before fracture;

[0047] b. Define a region of interest for the reference image and apply this region to all deformed images to calculate the displacement pixel distance in each deformed image;

[0048] c. Calculate the displacement pixel distance in each deformed image, convert the displacement data into actual physical quantity after pixel calibration, and then calculate the in-plane strain of the fabric;

[0049] d. Perform a visual display based on the calculation results to intuitively reflect the deformation of the fabric during the stretching process.

[0050] The above technical solution has the following beneficial effects:

[0051] 1. Automated processing and simplified operation: The use of automated video acquisition and image processing technology simplifies the operation process and reduces the possibility of human intervention. This systematic and standardized testing process improves test efficiency while reducing the risk of human error, allowing operators to complete complex testing tasks without deep professional knowledge, broadening the application scope of the equipment.

[0052] 2. Precision deformation measurement: Through high-resolution cameras and advanced computer vision technology, it is possible to accurately detect tiny deformations of fabrics, making the evaluation of fabric tensile properties more reliable and comprehensive.

[0053] 3. Multi-dimensional tensile strain detection and visualization: Through the improved image analysis method, the tensile strain, displacement and deformation characteristics of the fabric can be fully analyzed, including the three-dimensional morphology of the in-plane strain. This detailed analysis can provide a comprehensive performance evaluation of the fabric in practical applications.

[0054] 4. Stretch tester integrates visual recording: Improve the traditional stretch tester, expand the instrument storage space, integrate the camera bracket into the test equipment, and easily adjust the test distance and height during use. The visual recording system can capture every subtle change of the fabric during the stretching process and analyze it synchronously with the strain data, thus providing more comprehensive and three-dimensional information for fabric performance evaluation. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] The present invention will be further described below in conjunction with the accompanying drawings:

[0056] Figure 1 It is a flow chart of the visual intelligent detection method of the present invention;

[0057] Figure 2 It is a structural schematic diagram of a visual detection device for fabric tensile strain;

[0058] Figure 3 Flowchart for creating speckle patterns on the surface of fabric specimens.

[0059] in, Figure 2 The accompanying drawings are marked as: base 1, storage slot 2, adjustable camera bracket 3, computer 4, upper clamping device 5, moving track 6, light belt 7, stretching drive motor 8, moving device 9, drive switch 10, groove 11, lower clamping device 12. DETAILED DESCRIPTION

[0060] The device and method designed in the present invention include hardware and system parts, and are intended to perform intelligent tensile strain detection on various types of fabrics. The specific objectives include: (1) designing a visual detection device that is easy to fold, store, and disassemble; (2) realizing non-contact full-process measurement to ensure that the fabric is not damaged; (3) recording the appearance deformation of the fabric in real time, and quantifying it into external contour indicators such as elongation and contraction at the narrowest point, while calculating the Poisson's ratio to reflect the lateral deformation of the material and converting it into the tensile strain level of the fabric according to the standard; (4) outputting in real time the visual expression of the in-plane displacement of the fabric in the vertical and horizontal directions, as well as the visual expression of the in-plane strain distribution of the fabric in the vertical, horizontal, and shear directions, to achieve global refined characterization.

[0061] The technical solution of the present invention is described in detail below in conjunction with specific embodiments:

[0062] (1) Fabric sample pretreatment: Before the experiment, the fabric sample needs to be cut, speckle patterned, and balanced in a constant temperature and humidity environment for 24 hours to ensure its stability. The specific steps are as follows:

[0063] 1) Sample cutting: Cut the fabric into standard size according to relevant standards or user-specified size.

[0064] 2) Speckle production: Based on the sample specifications, a speckle pattern that meets the requirements of high contrast, randomness, non-directionality, and stability is automatically generated through computer simulation; an acrylic plate is used to perform blank processing based on the speckle position, and then a roller brush is used to dip ink to transfer the speckle pattern to the surface of the fabric sample to complete the speckle production;

[0065] 3) Temperature adjustment and debugging treatment: The fabric samples after the above treatment are placed in a constant temperature and humidity environment for 24 hours to stabilize the physical properties of the fabric samples and ensure the reliability and repeatability of the experimental data.

[0066] (2) Fabric stretching video acquisition: A detachable, foldable and storable visual detection device for fabric stretching strain is designed for fabric stretching video acquisition, aiming to facilitate instrument debugging and achieve full recording of the entire process of fabric stretching deformation. A detachable and foldable visual detection device is added to the traditional stretching instrument to record the deformation of the fabric in real time during the stretching process, providing an important basis for subsequent image segmentation, recognition and detection.

[0067] The device includes a base 1, an adjustable camera support 3, a movable track 6, an upper clamping device 5, a lower clamping device 12 and a computer 4. A storage slot 2 is provided on the right side of the base 1. The adjustable camera support 3 is installed in the storage slot 2. The storage slot 2 can store the folded adjustable camera support 3 and internalize it to prevent collision damage. The adjustable camera support 3 is used to install a camera. It is detachable, foldable, and adjustable in height and angle to adjust the shooting height and shooting angle of the camera, so as to achieve the purpose of completely recording the deformation picture of the fabric during the whole stretching process; the adjustable camera support 3 includes a connected horizontal bar and a vertical bar, both of which are foldable and detachable. The lower clamping device 12 is arranged on the base 1, facing upward, and is used to clamp the lower end of the fabric. The upper clamping device 5 is arranged on the movable track 6 of the base 1, and is used to clamp the upper end of the fabric. The upper clamping device 5 is moved to achieve the purpose of stretching the fabric.

[0068] The movable track 6 is connected with a stretching drive motor 8, a moving device 9 and a drive switch 10. The stretching drive motor 8 drives the moving device 9 to move on the movable track 6. The moving device 9 is connected with an upper clamping device 5 to drive the upper clamping device 5 to move on the movable track 6. The drive switch 10 is arranged on the base 1. The movable track 6 is also equipped with a light belt 7, which is used to provide sufficient light source.

[0069] The computer 4 is installed on the base 1 for real-time input and output of experimental data and visual capture of images; the base 1 is also provided with a groove 11 for placing a background plate to ensure that the background of the recorded image is clean and clear.

[0070] The collection process includes: adjusting the distance between the upper clamping device 5 and the lower clamping device 12 to clamp the fabric sample between the two; adjusting the height of the device and the distance between the camera and the fabric sample to ensure that the fabric deformation image during the entire stretching process is completely recorded; turning on the light strip 7 to provide sufficient light source, and installing the background board to prevent the background from affecting the shooting.

[0071] (3) Video sequence frame preprocessing: The collected fabric stretching videos are batch processed, images are extracted according to the time series, and the video sequence frame images are preprocessed, which specifically includes the following steps:

[0072] 1) ROI extraction: Through binarization processing and maximum bounding rectangle algorithm, the region of interest (ROI) where the fabric is located in each set of video frames is preliminarily determined to reduce computational redundancy.

[0073] 2) Image enhancement: The cropped region of interest is converted into a grayscale image, and adaptive histogram equalization is applied to enhance the contrast and improve the clarity of fabric details.

[0074] 3) Contour detection and mask creation: Binarize the enhanced image, identify the largest outer contour, and create a minimum bounding rectangle mask based on it for subsequent analysis.

[0075] (4) Extraction and calculation of deformation feature indicators: including real-time feature point extraction and matching and deformation calculation to obtain the elongation, contraction at the narrowest point and Poisson's ratio. The specific steps are as follows:

[0076] 1) Pixel calibration: Convert pixel values ​​to actual physical sizes to extract stress-strain core indicators, effective area length and width, deformation, Poisson’s ratio and other data.

[0077] 2) Real-time feature point extraction and matching: Extract key feature points from the first frame of the undeformed image, and use an improved image correlation analysis algorithm to compare the deformed image with the reference image to obtain the displacement field; at the same time, during the deformation image analysis process, use the trained model to perform real-time feature point extraction and matching to identify and track the feature points on the fabric surface to ensure high-precision displacement calculation.

[0078] 3) Deformation calculation: Calculate the deformation of each feature point, including elongation, contraction at the narrowest point, and Poisson’s ratio:

[0079] 1) Elongation calculation:

[0080] ΔL i =L i -L0

[0081] Where: L0 is the initial length of the effective area; Li is the length of the i-th interval frame image after stretching;

[0082] 2) Calculation of shrinkage at the narrowest point:

[0083] ΔW i =W i -W0

[0084] Where: W0 is the initial width, W i is the width after stretching;

[0085] 3) Poisson's ratio calculation:

[0086]

[0087] (5) Improvement of image correlation analysis algorithm: The improved image correlation analysis algorithm is applied, and the initial frame image is used as the reference image and other images are used as deformation images to analyze the in-plane dynamic deformation of the fabric during the stretching process. A deep learning model is introduced for feature matching to further improve the accuracy and speed of deformation detection. At the same time, a performance rating mechanism is added to objectively evaluate the mechanical properties of different fabrics. The deformation process is dynamically displayed through data visualization technology, so that researchers can understand the mechanical properties of the fabric more intuitively.

[0088] Using an integrated data acquisition and processing module, the step (3) and the step (4) are integrated into the module of the step (5), and the following operations are performed:

[0089] 1) Image registration: processing the acquired frame images according to step (3) and step (4), taking the first frame image as a reference image and the remaining frame images as deformation images, so as to analyze the in-plane deformation of the fabric during the stretching process;

[0090] a. Select the first frame image at the beginning of stretching as the reference image, and the last frame image as the image before fracture;

[0091] b. Define a region of interest for the reference image and apply this region to all deformed images to calculate the displacement pixel distance in each deformed image;

[0092] c. Calculate the displacement pixel distance in each deformed image, convert the displacement data into actual physical quantity after pixel calibration, and then calculate the in-plane strain of the fabric;

[0093] d. Perform a visual display based on the calculation results to intuitively reflect the deformation of the fabric during the stretching process.

[0094] 2) Correlation calculation: The zero mean normalized least square distance correlation function (ZNSSD) is used as the correlation criterion, and its calculation method is:

[0095]

[0096] Where: f(x i ,y i ), p(x′ i ,y′ i ) are the grayscale values ​​of the pixels in the reference sub-region and the current sub-region respectively; f m 、p m are the average gray values ​​corresponding to the reference sub-area and the deformed sub-area respectively; x i ,y iare the horizontal and vertical coordinates of the pixels in the corresponding deformed sub-area in the whole image; i and j are the numbers of the pixels in the sub-area; σ1 and σ2 are the grayscale standard deviations of the reference sub-area and the current sub-area, indicating the degree of change or the breadth of the distribution of the grayscale value of the image; M and N are the height and width of the image, respectively; MN is the number of pixels in the image; f and p are the reference sub-area and the current sub-area, respectively, with a size of M×N; It means to sum the positions of all pixels in the image;

[0097] 3) Iterative optimization and model training: Use the Newton-Raphson iteration method to optimize calculations, and use labeled data sets to train deep learning models to improve the accuracy of feature matching.

[0098] 4) Strain analysis and performance rating: Calculate the strain field of the fabric in real time, generate a strain distribution diagram, and rate the mechanical properties based on indicators such as stress-strain curve and Poisson's ratio.

[0099] 5) Data visualization technology: Using data visualization technology, the deformation process is dynamically displayed, animation effects are generated, the deformation state of the fabric during the stretching process is clearly presented, and a visualization report is generated, including key performance indicators such as stress-strain curve and Poisson's ratio.

[0100] 6) System optimization and adaptability: Adaptive learning function can adjust processing algorithm parameters according to real-time data feedback to further improve detection accuracy and efficiency.

[0101] The above are only specific embodiments of the present invention, but the technical features of the present invention are not limited thereto. Any simple changes, equivalent replacements or modifications made based on the present invention to solve basically the same technical problems and achieve basically the same technical effects are all included in the protection scope of the present invention.

Claims

1. A visual intelligent detection method for fabric tensile strain: characterized by The steps include: (1) Fabric stretching video acquisition; (2) Video sequence frame preprocessing: The collected fabric stretching videos are batch processed, images are extracted according to the time series, and the video sequence frame images are preprocessed; (3) Extraction and calculation of deformation feature indicators: including real-time feature point extraction and matching and deformation calculation to obtain elongation, contraction at the narrowest point and Poisson's ratio; (4) Improvement of image correlation analysis algorithm: In the tensile test, the improved image correlation analysis algorithm is applied, and the initial frame image is used as the reference image and other images are used as deformation images to analyze the in-plane dynamic deformation of the fabric during the stretching process. A deep learning model is introduced for feature matching to improve the accuracy and speed of deformation detection. At the same time, a performance rating mechanism is added to objectively evaluate the mechanical properties of different fabrics, and the deformation process is dynamically displayed through data visualization technology.

2. The visual intelligent detection method of fabric tensile strain according to claim 1 is characterized by: Fabric samples need to be pre-treated, including: 1) Sample cutting: Cut the fabric into standard size according to relevant standards or user-specified size. 2) Speckle production: Based on the sample specifications, a speckle pattern that meets the requirements of high contrast, randomness, non-directionality, and stability is automatically generated through computer simulation; an acrylic plate is used to perform blank processing based on the speckle position, and then a roller brush is used to dip ink to transfer the speckle pattern to the surface of the fabric sample to complete the speckle production; 3) Temperature adjustment and debugging treatment: The fabric sample after the above treatment is placed in a constant temperature and humidity environment for 24 hours to adjust the temperature and humidity to stabilize the physical properties of the fabric sample.

3. The visual intelligent detection method of fabric tensile strain according to claim 1 is characterized by: The step (1) uses a visual detection device for fabric tensile strain to collect fabric stretching video, the device includes a base, an adjustable camera bracket, a movable track, an upper clamping device, a lower clamping device and a computer, the adjustable camera bracket is used to install a camera, and is detachable, foldable, and adjustable in height and angle; the lower clamping device and the upper clamping device respectively clamp the two ends of the fabric for stretching the fabric; the movable track is connected to a stretching drive motor, a movable device and a drive switch, the stretching drive motor drives the movable device to move on the movable track, and the movable device is connected to the upper clamping device for driving the upper clamping device to move on the movable track; The computer is used for real-time input and output of experimental data and visual capture images; The collection process includes: adjusting the distance between the upper clamping device and the lower clamping device to clamp the fabric sample between the upper clamping device and the lower clamping device; Adjust the height of the device and the distance between the camera and the fabric sample to ensure that the fabric deformation image during the entire stretching process is completely recorded.

4. The visual intelligent detection method of fabric tensile strain according to claim 1 is characterized by: A storage slot is provided at the position where the adjustable camera bracket is installed on the base, and the storage slot is used to store the adjustable camera bracket; a groove is also provided on the base, and the groove is used to place a background plate; a light strip is also installed on the movable track, and the light strip is used to provide a light source.

5. The visual intelligent detection method of fabric tensile strain according to claim 1 is characterized by: The step (2), video sequence frame image preprocessing comprises: 1) Region of interest extraction: Through binarization processing and maximum bounding rectangle algorithm, the region of interest where the fabric is located in each set of video frames is preliminarily determined to reduce computational redundancy. 2) Image enhancement: Convert the cropped region of interest to a grayscale image and apply adaptive histogram equalization to enhance the contrast; 3) Contour detection and mask creation: Binarize the enhanced image, identify the largest outer contour, and create a minimum bounding rectangle mask based on it.

6. The visual intelligent detection method of fabric tensile strain according to claim 1 is characterized by: The step (3) also includes pixel calibration: converting the pixel value into the actual physical size in advance.

7. The visual intelligent detection method of fabric tensile strain according to claim 1 is characterized by: The step (3) is real-time feature point extraction and matching: extracting key feature points from the first frame image that has not been deformed, and using an improved image correlation analysis algorithm to compare the deformed image with the reference image to obtain a displacement field; at the same time, during the deformation image analysis process, using the trained model to perform real-time feature point extraction and matching to identify and track feature points on the fabric surface.

8. The visual intelligent detection method of fabric tensile strain according to claim 7 is characterized by: The step (3), deformation calculation: calculate the deformation of each feature point, including the elongation, the contraction at the narrowest point, and the Poisson's ratio: 1) Elongation calculation: ΔL i =L i -L0 Where: L0 is the initial length of the effective area; L i is the length of the i-th interval frame image after stretching; 2) Calculation of shrinkage at the narrowest point: ΔW i =In i -W0 Where: W0 is the initial width, Wi is the width after stretching; 3) Poisson's ratio calculation:

9. The visual intelligent detection method of fabric tensile strain according to claim 1, characterized in that: The step (4) uses an integrated data acquisition and processing module to integrate the step (2) and the step (3) into the module of the step (4), and performs the following operations: 1) Image registration: processing the acquired frame images according to step (3) and step (4), taking the first frame image as a reference image and the remaining frame images as deformation images, so as to analyze the in-plane deformation of the fabric during the stretching process; 2) Correlation calculation: The zero-mean normalized least square distance correlation function ZNSSD is used as the correlation criterion, and its calculation method is: Where: f(x i ,y i ), p(x′ i ,y′ i ) are the grayscale values ​​of the pixels in the reference sub-region and the current sub-region respectively; f m 、p m are the average gray values ​​corresponding to the reference sub-area and the deformed sub-area respectively; x i ,y i are the horizontal and vertical coordinate values ​​of the pixels in the corresponding deformed sub-area in the whole image; i and j are the numbers of the pixels in the sub-area; σ1, σ 2 are the grayscale standard deviations of the reference sub-area and the current sub-area, respectively, indicating the degree of change or the breadth of distribution of the grayscale value of the image; M and N are the height and width of the image respectively; MN is the number of pixels in the image; f and p are the reference sub-region and the current sub-region respectively, with a size of M×N; It means to sum the positions of all pixels in the image; 3) Iterative optimization and model training: Use the Newton-Raphson iteration method to optimize the calculation, and use the labeled data set to train the deep learning model; 4) Strain analysis and performance rating: Real-time calculation of the strain field of the fabric, generation of strain distribution diagram, and mechanical performance rating based on stress-strain curve and Poisson’s ratio; 5) Data visualization technology: Using data visualization technology, the deformation process is dynamically displayed, animation effects are generated, the deformation state of the fabric during the stretching process is clearly presented, and a visualization report is generated, including stress-strain curves and Poisson's ratio; 6) System optimization and adaptability: Adaptive learning function adjusts processing algorithm parameters based on real-time data feedback.

10. A visual intelligent detection method for fabric tensile strain according to claim 9, characterized in that: The specific steps of step 1) image registration include: a. Select the first frame image at the beginning of stretching as the reference image, and the last frame image as the image before fracture; b. Define a region of interest for the reference image and apply this region to all deformed images to calculate the displacement pixel distance in each deformed image; c. Calculate the displacement pixel distance in each deformed image, convert the displacement data into actual physical quantity after pixel calibration, and then calculate the in-plane strain of the fabric; d. Perform a visual display based on the calculation results to intuitively reflect the deformation of the fabric during the stretching process.

Images