Fabric weave structure intelligent detection method and related device

Abstract

The embodiment of the invention provides a fabric weave structure intelligent detection method and a related device, and the method comprises the steps: obtaining a to-be-detected fabric image from a terminal, the terminal being a Web terminal or a mobile terminal, and the fabric image being obtained through the shooting of a portable microscope; preprocessing the fabric image to obtain a target fabric image; the target fabric image is input to a pre-trained deep learning model, a fabric weave structure result is output, the deep learning model is constructed based on a target detection algorithm and is suitable for weave structures of various woven fabrics, knitted fabrics and non-woven fabrics, and the fabric weave structure result is obtained. The deep learning model comprises a target detection layer, a feature extraction network and a classifier. On the basis of the method, the organization structures of various textile fabrics can be quickly and accurately recognized, the detection efficiency and accuracy are improved, the manual detection cost is saved, the labor intensity is reduced, and automatic and intelligent recognition of the fabric organization structures is achieved.

Description

Technical Field

[0001] The embodiments of the present invention relate to the technical field of textile testing and analysis, and in particular to a method and related device for intelligently detecting fabric structure. Background Art

[0002] The detection and identification of traditional fabric structure relies on manual visual inspection or the use of complex laboratory equipment. This method is not only time-consuming and labor-intensive, but also easily affected by subjective factors, and it is difficult to meet the modern textile industry's needs for efficient and accurate detection. Summary of the invention

[0003] The embodiment of the present invention provides a fabric structure intelligent detection method and related devices, which can quickly and accurately identify a variety of textile structures, not only improving the detection efficiency and accuracy, but also saving the cost and labor intensity of manual detection.

[0004] In a first aspect, an embodiment of the present invention provides a method for intelligently detecting a fabric structure, which is applied to a server, and the method includes:

[0005] Acquire a fabric image to be detected from a terminal, wherein the terminal is a Web terminal or a mobile terminal, and the fabric image is obtained by taking a photo with a portable microscope;

[0006] Preprocessing the fabric image to obtain a target fabric image;

[0007] The target fabric image is input into a pre-trained deep learning model, and the fabric structure result is output, wherein the deep learning model is constructed based on a target detection algorithm, and the deep learning model includes a target detection layer, a feature extraction network and a classifier. The deep learning model outputs the fabric structure result according to the multi-dimensional information of the target fabric image, and the multi-dimensional information includes the weaving rules of the yarns or fibers in the target fabric.

[0008] In some embodiments, the deep learning model training method includes:

[0009] Acquire a data set, wherein the data set includes double-sided images of fabrics of different types and different tissue structures and corresponding weaving rules of yarns or fibers, and the double-sided images of the fabrics include a front side image of the fabric and a back side image of the fabric;

[0010] Annotating the target area position, category information, and weaving regularity of yarns or fibers in the double-sided fabric image to obtain annotation data;

[0011] Build a lightweight convolutional neural network;

[0012] Based on the labeled data, an optimization algorithm is used to perform model training on the lightweight convolutional neural network to adjust target model parameters, wherein the model parameters include weights of features at each level in the channel domain and the spatial domain;

[0013] The trained deep learning model is obtained according to the target model parameters.

[0014] In some embodiments, the preprocessing of the fabric image to obtain a target fabric image includes:

[0015] Using Gaussian filtering or median filtering to remove image noise of the fabric image to obtain a first fabric image;

[0016] Using a contrast enhancement algorithm to adjust the image brightness distribution of the first fabric image to obtain a second fabric image;

[0017] adjusting the image color of the second fabric image using a color correction algorithm to obtain a third fabric image;

[0018] Using a threshold-based segmentation algorithm, a region-based segmentation algorithm, or an edge-based segmentation algorithm to segment different tissue regions of the third fabric image to obtain segmented tissue regions;

[0019] Performing image enhancement processing on the segmented tissue area to obtain a target fabric image, wherein the image enhancement processing includes sharpening filtering and edge detection;

[0020] Fourier transform and eigenvalue extraction methods are used on the target fabric image to obtain the weaving pattern of the yarns or fibers in the target fabric. The weaving pattern includes the weaving trajectory and direction of the yarns or fibers, which are left-right crossing along the warp, up-and-down crossing along the weft, vertical floating and sinking along the warp, parallel floating and sinking along the weft, and random and disordered multi-directional crossing.

[0021] In some embodiments, the step of inputting the target fabric image into a pre-trained deep learning model and outputting a fabric structure result includes:

[0022] Using the target detection layer to perform target detection on the target fabric image, and extracting the target area of ​​the fabric structure;

[0023] The feature extraction network is used to extract multi-dimensional information from the target area, wherein the multi-dimensional information includes shape features, texture features, color features, and weaving rules of yarns or fibers;

[0024] Generate a feature vector according to the multi-dimensional information;

[0025] The classifier is used to classify the feature vector to obtain the fabric structure result.

[0026] In some embodiments, the method further comprises:

[0027] The fabric images are classified and stored according to date and sample number.

[0028] In a second aspect, an embodiment of the present invention provides a method for intelligently detecting a fabric structure, which is applied to a terminal, wherein the terminal is a Web terminal or a mobile terminal, and the method includes:

[0029] Acquire a fabric image to be detected, wherein the fabric image is obtained by photographing through a portable microscope;

[0030] Sending the fabric image to a server so that the server executes the fabric weave structure intelligent detection method as described in the first aspect to generate a fabric weave structure result;

[0031] The fabric weave structure result is received from the server.

[0032] In some embodiments, the method further comprises:

[0033] Display the fabric structure result;

[0034] A test report is generated according to the fabric structure result and user demand information, and the test report includes test time, test result and image comparison information.

[0035] In the third aspect, an embodiment of the present invention further provides a fabric structure intelligent detection system, the system comprising a server and a terminal, the terminal and the server interacting with each other through an API interface information, the server executing the fabric structure intelligent detection method as described in the first aspect, and the terminal executing the fabric structure intelligent detection method as described in the second aspect.

[0036] In a fourth aspect, an embodiment of the present invention further provides an electronic device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the intelligent detection method for fabric structure as described in the first aspect, or the intelligent detection method for fabric structure as described in the second aspect is implemented.

[0037] In a fifth aspect, an embodiment of the present invention further provides a computer-readable storage medium storing computer-executable instructions, wherein the computer-executable instructions are used to execute the intelligent detection method for fabric structure as described in the first aspect, or the intelligent detection method for fabric structure as described in the second aspect.

[0038] According to the intelligent detection method and related device of fabric structure provided by the embodiment of the present invention, the intelligent detection method of fabric structure includes: obtaining a fabric image to be detected from a terminal, wherein the terminal is a Web terminal or a mobile terminal, and the fabric image is obtained by photographing with a portable microscope; preprocessing the fabric image to obtain a target fabric image; inputting the target fabric image into a pre-trained deep learning model, and outputting the fabric structure result, wherein the deep learning model is constructed based on a target detection algorithm, and the deep learning model includes a target detection layer, a feature extraction network and a classifier, and the deep learning model outputs the fabric structure result according to the multi-dimensional information of the target fabric image, and the multi-dimensional information includes the weaving law of the yarn or fiber in the target fabric. The fabric image is obtained by photographing the fabric with a portable microscope, and the fabric image is uploaded to the server through the terminal for image processing and deep learning recognition. In the deep learning recognition stage, the pre-processed target fabric image is subjected to target detection by the target detection layer, and the target area of ​​the structure is extracted, and then the multi-dimensional information such as shape features, texture features, color features, and weaving law of yarn or fiber are extracted from the target area by using the feature extraction network to form a feature vector. The classifier is then used to classify the feature vectors to achieve intelligent detection and identification of the fabric structure. Based on this, the embodiment of the present invention can quickly and accurately identify a variety of textile structures, which not only improves the detection efficiency and accuracy, but also saves the cost and labor intensity of manual detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 is a flow chart of a method for intelligently detecting a fabric structure applied to a server provided by an embodiment of the present invention;

[0040] Figure 2A is a flow chart of a training method for a deep learning model provided by an embodiment of the present invention;

[0041] Figure 2B is a schematic diagram of a dual-branch network architecture provided by an embodiment of the present invention;

[0042] Figure 3 yes Figure 1 Flow chart of step S102;

[0043] Figure 4 is a flow chart of step S103 in FIG2 ;

[0044] Figure 5 is a flow chart of a method for intelligently detecting a fabric structure of an application terminal provided by an embodiment of the present invention;

[0045] Figure 6 is a flowchart of steps S601 to S602;

[0046] Figure 7 is a schematic diagram of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0047] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0048] It should be noted that, although the functional modules are divided in the device schematic diagram and the logical order is shown in the flowchart, in some cases, the steps shown or described may be performed in a different order than the module division in the device or the order in the flowchart. The terms "first", "second", etc. in the specification and claims and the following drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence.

[0049] In the embodiments of the present invention, the words "further", "exemplarily" or "optionally" are used to indicate examples, illustrations or descriptions, and should not be interpreted as being more preferred or more advantageous than other embodiments or designs. The use of the words "further", "exemplarily" or "optionally" is intended to present related concepts in a specific way.

[0050] In order to more conveniently describe the working principle of the embodiment of the present invention later, an introduction to relevant technical scenarios is first given below.

[0051] The detection and identification of traditional fabric structure relies on manual visual inspection or the use of complex laboratory equipment. This method is not only time-consuming and labor-intensive, but also easily affected by subjective factors, and it is difficult to meet the modern textile industry's needs for efficient and accurate detection.

[0052] Based on this, the present invention provides a fabric structure intelligent detection method and related devices. The fabric structure intelligent detection method includes: obtaining a fabric image to be detected from a terminal, wherein the terminal is a Web terminal or a mobile terminal, and the fabric image is obtained by photographing with a portable microscope; preprocessing the fabric image to obtain a target fabric image; inputting the target fabric image into a pre-trained deep learning model, and outputting the fabric structure result, wherein the deep learning model is constructed based on a target detection algorithm, and the deep learning model includes a target detection layer, a feature extraction network and a classifier. The deep learning model outputs the fabric structure result according to the multi-dimensional information of the target fabric image, and the multi-dimensional information includes the weaving law of the yarn or fiber in the target fabric. The fabric image is obtained by photographing the fabric with a portable microscope, and the fabric image is uploaded to the server through the terminal for image processing and deep learning recognition. In the deep learning recognition stage, the pre-processed target fabric image is subjected to target detection by the target detection layer, and the target area of ​​the structure is extracted, and then the multi-dimensional information such as shape features, texture features, color features, and weaving laws of yarns or fibers are extracted from the target area by using the feature extraction network to form a feature vector. The classifier is then used to classify the feature vectors to achieve intelligent detection and recognition of the fabric structure. Based on this, the embodiment of the present invention can quickly and accurately identify the structure of various textile fabrics, which not only improves the detection efficiency and accuracy, but also saves the cost of manual detection and reduces labor intensity, and realizes automatic and intelligent recognition of fabric structure.

[0053] The embodiments of the present invention are further described below in conjunction with the accompanying drawings.

[0054] like Figure 1 As shown, Figure 1 It is a flow chart of a method for intelligently detecting a fabric structure provided by an embodiment of the present invention. The method for intelligently detecting a fabric structure is applied to a server. The method may include but is not limited to steps S101 to S103.

[0055] Step S101, obtaining a fabric image to be detected from a terminal, wherein the terminal is a Web terminal or a mobile terminal, and the fabric image is obtained by taking a photo with a portable microscope;

[0056] Step S102, preprocessing the fabric image to obtain a target fabric image;

[0057] Step S103, input the target fabric image into a pre-trained deep learning model, and output the fabric structure result, wherein the deep learning model is constructed based on the target detection algorithm, and the deep learning model includes a target detection layer, a feature extraction network and a classifier. The deep learning model outputs the fabric structure result according to the multi-dimensional information of the target fabric image, and the multi-dimensional information includes the weaving rules of the yarns or fibers in the target fabric.

[0058] It is understandable that the fabric image to be tested can be obtained from the Web or mobile terminal, and the fabric image can be photographed by a portable microscope, and the photographed fabric image can be transmitted to the Web or mobile terminal. A portable microscope is used as a data acquisition device. The portable microscope has an adjustable magnification and high-definition imaging capability, which can clearly capture the details of the fabric structure. Design a special fixture or platform to ensure that the fabric sample remains stable during the inspection process to avoid image blur caused by movement or deformation. The magnification, focus and other parameters of the microscope can also be controlled by the data acquisition software, and image acquisition can be automatically or manually triggered to save the acquired fabric image to local or cloud storage.

[0059] It is understandable that the collected fabric images can be preprocessed, including denoising, contrast enhancement, color correction, etc., to improve image quality and lay the foundation for subsequent processing. Image segmentation technology is used to segment different tissue areas in the fabric image to facilitate subsequent target detection and feature extraction. Image enhancement algorithms, such as histogram equalization and sharpening filtering, can be used to further improve the image visualization effect and recognition accuracy.

[0060] It is understandable that a deep learning model based on advanced target detection algorithms such as YOLOv9 can be constructed to identify the weave structure in fabric images, such as the warp arrangement and shape, weft arrangement and shape, coils, etc. A feature extraction network is designed to extract multi-dimensional information such as fiber and yarn distribution patterns and shape characteristics, texture characteristics, color characteristics, etc. from the detected target area to form a feature vector. The feature vectors are classified using a machine learning classifier or a deep learning classifier to achieve intelligent recognition of the fabric weave structure. At the same time, the model parameters and structure are continuously optimized to improve recognition accuracy and generalization ability.

[0061] It is understandable that the weaving rules of the yarns or fibers in the target fabric, i.e., the weaving tracks and orientation rules of the yarns or fibers in the target fabric, include but are not limited to left-right crossing along the warp, up-and-down crossing along the weft, vertical floating and sinking along the warp, parallel floating and sinking along the weft, and random and disordered multi-directional crossing. The fabric structure intelligent detection method of the present invention is suitable for automatic and intelligent identification of the structure of various woven fabrics, knitted fabrics and non-woven fabrics.

[0062] Based on this, a fabric image is obtained by photographing the fabric through a portable microscope, and the fabric image is uploaded to the server through the Web or mobile terminal for image processing and deep learning recognition. In the deep learning recognition stage, the pre-processed target fabric image is subjected to target detection through the target detection layer to extract the target area of ​​the tissue structure, and then the shape feature, texture feature, color feature and other multi-dimensional information are extracted from the target area using the feature extraction network to form a feature vector. The feature vector is then classified using a classifier to achieve intelligent detection and recognition of the fabric structure. Based on this, the embodiment of the present invention can quickly and accurately identify the tissue structure of various textile fabrics, which not only improves the detection efficiency and accuracy, but also saves the cost of manual detection and reduces labor intensity, and realizes the automation and intelligence of the entire process from data collection to result display.

[0063] like Figure 2A As shown, the training method of the deep learning model may include but is not limited to steps S201 to S205.

[0064] Step S201, obtaining a data set, wherein the data set includes double-sided images of fabrics of different types and different tissue structures and corresponding weaving rules of yarns or fibers, and the double-sided images of the fabrics include a front image of the fabric and a back image of the fabric;

[0065] Step S202, marking the target area position, category information, and weaving regularity of yarns or fibers in the double-sided fabric image to obtain marked data;

[0066] Step S203, constructing a lightweight convolutional neural network;

[0067] Step S204, using an optimization algorithm to perform model training on the lightweight convolutional neural network based on the labeled data to adjust the target model parameters, where the model parameters include the weights of each level of features in the channel domain and the spatial domain;

[0068] Step S205, obtaining a trained deep learning model according to the target model parameters.

[0069] It is understandable that for model construction, a deep learning model can be constructed based on advanced target detection algorithms such as YOLOv9. The model can include a feature extraction network, a target detection layer, and a classifier. During the construction process, the network structure and parameters need to be adjusted according to the characteristics of the fabric structure.

[0070] It is understandable that for the preparation of the data set, in order to ensure the generalization ability of the model, the present invention has been carefully designed in the construction of the data set. The collected samples cover a variety of materials (such as pure cotton, spandex, polyester, etc.) and fabric types with a variety of surface density ranges, and images of the front and back sides of each fabric are taken. During the shooting process, the diversity of the samples is increased by changing the light intensity, shooting angle, etc. to improve the robustness of the model. A large amount of image data containing fabrics of different types and different organizational structures is collected and annotated. The annotation should include information such as the location of the target area (such as a bounding box), category (such as warp, weft, etc.). Use these data to train a deep learning model.

[0071] It is understandable that if Figure 2B As shown, the present invention proposes an innovative dual-branch network architecture for processing double-sided images of fabrics. This architecture can extract feature information on both sides of the fabric in parallel, overcomes the limitation of traditional methods that only rely on single-sided images for recognition, and inputs the front and back sides of a fabric structure into the network as a sample for training. By considering the features of both sides of the fabric at the same time, the accuracy and robustness of the tissue structure classification are significantly improved. In order to reduce the consumption of computing resources, the present invention adopts a lightweight convolutional neural network (such as GhostNet, MobileNet, ShuffleNet) as the basic model. While ensuring high performance, this model greatly reduces the number of parameters and computational complexity, making it more suitable for deployment in practical applications, especially in resource-constrained environments. In the feature extraction stage, an attention mechanism module is introduced to correct the weights of features at each level in the channel domain and the spatial domain. This mechanism can enhance the network's ability to capture key features and improve the feature distinction between different tissues, so that it can still maintain a high classification accuracy under interference such as angle rotation, scale change, and changes in lighting conditions.

[0072] It is understandable that for model training and optimization, appropriate optimization algorithms (such as Adam, SGD, etc.) can be used to train the model and continuously adjust the model parameters to improve recognition accuracy and generalization ability. At the same time, cross-validation and other technologies are used to evaluate model performance to avoid overfitting.

[0073] It is understandable that for feature extraction and classification, the trained deep learning model can be used to perform target detection on the preprocessed fabric image to extract the target area of ​​the weave structure. Then, multi-dimensional information such as shape features, texture features, and color features are extracted from the target area to form a feature vector. Finally, the feature vector is classified using a machine learning classifier or a deep learning classifier to achieve intelligent recognition of the fabric weave structure.

[0074] In one embodiment, if Figure 3As shown, step S102 may include but is not limited to steps S301 to S306.

[0075] Step S301, using Gaussian filtering or median filtering to remove image noise of the fabric image to obtain a first fabric image;

[0076] Step S302, adjusting the image brightness distribution of the first fabric image by using a contrast enhancement algorithm to obtain a second fabric image;

[0077] Step S303, adjusting the image color of the second fabric image using a color correction algorithm to obtain a third fabric image;

[0078] Step S304, using a threshold-based segmentation algorithm, a region-based segmentation algorithm, or an edge-based segmentation algorithm to segment different tissue regions of the third fabric image to obtain segmented tissue regions;

[0079] Step S305, performing image enhancement processing on the segmented tissue area to obtain a target fabric image, wherein the image enhancement processing includes sharpening filtering and edge detection;

[0080] Step S306, using Fourier transform and eigenvalue extraction methods on the target fabric image to obtain the weaving pattern of the yarns or fibers in the target fabric, the weaving pattern including the weaving trajectory and direction of the yarns or fibers crossing left and right along the warp direction, crossing up and down along the weft direction, vertically floating and sinking along the warp direction, parallelly floating and sinking along the weft direction, and randomly and disorderly crossing in multiple directions.

[0081] It is understandable that a variety of image processing algorithms can be used to pre-process the collected fabric images, including but not limited to: Gaussian filtering or median filtering to remove image noise. Contrast enhancement algorithms (such as histogram equalization) to improve image brightness distribution. Color correction algorithms to adjust image colors to ensure that the colors truly reflect the original appearance of the fabric.

[0082] It is understood that for image segmentation, different tissue regions in the fabric image can be segmented using algorithms such as threshold-based segmentation, region-based segmentation, or edge-based segmentation. The segmentation result should be as accurate as possible to reduce interference with subsequent target detection.

[0083] It is understandable that for image enhancement, further image enhancement processing, such as sharpening filtering, edge detection, etc., can be performed on the segmented tissue area to improve the clarity and recognition of the target area.

[0084] In one embodiment, if Figure 4 As shown, step S103 may include but is not limited to steps S401 to S404.

[0085] Step S401, using the target detection layer to perform target detection on the target fabric image, and extracting the target area of ​​the fabric structure;

[0086] Step S402, extracting multi-dimensional information from the target area using a feature extraction network, the multi-dimensional information including shape features, texture features, color features, and weaving rules of yarns or fibers;

[0087] Step S403, generating a feature vector according to the multi-dimensional information;

[0088] Step S404: classify the feature vector using a classifier to obtain a fabric structure result.

[0089] It is understandable that the target detection layer of the trained deep learning model can be used to perform target detection on the preprocessed target fabric image to extract the target area of ​​the tissue structure. Then, multi-dimensional information such as shape features, texture features, and color features are extracted from the target area to form a feature vector. Finally, the feature vector is classified using a classifier to achieve intelligent recognition of the fabric structure, wherein the classifier includes but is not limited to a machine learning classifier or a deep learning classifier.

[0090] like Figure 5 As shown, Figure 5 It is a flowchart of a method for intelligent detection of fabric structure provided by an embodiment of the present invention. The method for intelligent detection of fabric structure is applied to a terminal, which is a Web terminal or a mobile terminal. The method may include but is not limited to steps S501 to S503.

[0091] Step S501, obtaining a fabric image to be detected, wherein the fabric image is obtained by photographing with a portable microscope;

[0092] Step S502, sending the fabric image to the server, so that the server executes the above-mentioned fabric weave structure intelligent detection method to generate a fabric weave structure result;

[0093] Step S503: receiving the fabric structure result from the server.

[0094] It is understandable that fabric images can be captured by a portable microscope, and the captured fabric images can be transmitted to a web or mobile terminal. A portable microscope is used as a data acquisition device. The portable microscope has adjustable magnification and high-definition imaging capabilities, which can clearly capture the details of the fabric structure. Design a dedicated fixture or platform to ensure that the fabric sample remains stable during the inspection process to avoid image blur caused by movement or deformation. The magnification, focus and other parameters of the microscope can also be controlled through the data acquisition software, and image acquisition can be automatically or manually triggered to save the acquired fabric image to local or cloud storage.

[0095] Understandably, for equipment selection, you can choose a high-resolution, portable microscope, such as a microscope equipped with a digital camera, to ensure that it can be magnified enough to clearly observe the fabric structure. At the same time, ensure that the microscope has an adjustable focus and light source to accommodate fabrics of different materials and colors.

[0096] It is understood that for the fixture design, a fixture suitable for fabric samples of various sizes and shapes can be designed and manufactured, and the fixture should be able to firmly clamp the fabric to prevent movement or deformation during the test. The surface of the fixture can be paved with anti-slip material to increase the friction with the fabric.

[0097] It is understandable that a user-friendly data acquisition software can be developed, which should have the following functions: Real-time display of fabric images under the microscope, supporting operations such as zooming and panning. Set the microscope's magnification, focal length, light source brightness and other parameters, and save them as preset configurations for subsequent use. Provide the function of automatically or manually triggering image acquisition, supporting continuous shooting and single shooting. Automatically save the collected images to a local storage device or cloud server, and support classification storage by date, sample number, etc.

[0098] It is understandable that the web or mobile terminal has a user interaction interface, which can provide an intuitive operation process and a friendly user experience. The web or mobile terminal has an image upload function, allowing users to upload the fabric image to be detected by taking a photo or selecting an existing picture. The web or mobile terminal is integrated with an API interface for docking with the server, which transmits the image uploaded by the user to the server for intelligent recognition, and feeds back the recognized fabric structure results to the user in real time.

[0099] It is understandable that the server performs image processing and deep learning recognition on the fabric image received from the Web or mobile terminal. In the deep learning recognition stage, the target detection layer performs target detection on the pre-processed target fabric image to extract the target area of ​​the structure, and then uses the feature extraction network to extract multi-dimensional information such as shape features, texture features, and color features from the target area to form a feature vector. The classifier is then used to classify the feature vector to achieve intelligent detection and recognition of the fabric structure.

[0100] like Figure 6 As shown, the fabric structure intelligent detection method may also include but is not limited to steps S601 to S602.

[0101] Step S601, displaying the fabric structure result;

[0102] Step S602: Generate a test report based on the fabric structure result and user requirement information, where the test report includes test time, test result and image comparison information.

[0103] It is understandable that the web or mobile terminal can provide a visual display of the recognition results, such as marking the detected organizational structure, displaying the classification results, etc., to facilitate user understanding and analysis. The web or mobile terminal also has a report generation function, allowing users to generate a test report according to their needs, including information such as test time, test results, image comparison, etc. The generated test report can support download or print functions, which is convenient for users to archive or share.

[0104] In addition, an embodiment of the present invention also discloses a fabric structure intelligent detection system, which includes a server and a terminal. The terminal and the server interact with each other through an API interface. The server executes the fabric structure intelligent detection method as described above, and the terminal executes the fabric structure intelligent detection method as described above.

[0105] It is understandable that the entire fabric structure intelligent detection system achieves rapid and accurate identification of various textile structures by integrating portable microscopes, deep learning algorithms and Web / mobile application technologies. The system not only improves detection efficiency and accuracy, reduces the cost and labor intensity of manual detection, but also provides users with convenient and efficient detection tools. At the same time, the scalability and customizability of the system also provide strong support for the further intelligent development of the textile industry.

[0106] It can be understood that the entire system integrates the data acquisition module, image processing module, deep learning recognition module and Web / mobile application interaction module to form a complete fabric structure intelligent detection system, ensuring smooth data transmission, fast processing speed and high accuracy between modules.

[0107] It can be understood that the specific implementation of the data acquisition module is as follows:

[0108] Equipment selection: Choose a high-resolution, portable microscope, such as one equipped with a digital camera, that can be magnified enough to clearly observe the fabric structure. Also, make sure the microscope has an adjustable focus and light source to accommodate fabrics of different materials and colors.

[0109] Fixture design: Design and manufacture a fixture suitable for fabric samples of various sizes and shapes. The fixture should be able to firmly clamp the fabric to prevent movement or deformation during testing. The surface of the fixture can be paved with non-slip material to increase friction with the fabric.

[0110] Data acquisition software: Develop a user-friendly data acquisition software that has the following functions: Real-time display of fabric images under the microscope, supporting operations such as zooming and panning. Set the microscope's magnification, focal length, light source brightness and other parameters, and save them as preset configurations for subsequent use. Provide automatic or manual triggering of image acquisition, supporting continuous shooting and single shooting. Automatically save the acquired images to a local storage device or cloud server, supporting classification storage by date, sample number, etc.

[0111] It can be understood that the specific implementation of the image processing module is as follows:

[0112] Preprocessing algorithm: Use a variety of image processing algorithms to preprocess the collected fabric images, including but not limited to: Gaussian filtering or median filtering to remove image noise. Contrast enhancement algorithms (such as histogram equalization) to improve image brightness distribution. Color correction algorithms to adjust image colors to ensure that the colors truly reflect the original appearance of the fabric.

[0113] Image segmentation: Different tissue regions in the fabric image are segmented using algorithms such as threshold-based segmentation, region-based segmentation, or edge-based segmentation. The segmentation results should be as accurate as possible to reduce interference with subsequent target detection.

[0114] Image enhancement: Further image enhancement processing is performed on the segmented tissue area, such as sharpening filtering, edge detection, etc., to improve the clarity and recognition of the target area.

[0115] It can be understood that the specific implementation of the deep learning recognition module is as follows:

[0116] Model construction: Based on advanced target detection algorithms such as YOLOv9, a deep learning model is constructed. The model should include feature extraction network, target detection layer, and classification layer. During the construction process, the network structure and parameters need to be adjusted according to the characteristics of the fabric structure.

[0117] Dataset preparation: In order to ensure the generalization ability of the model, the present invention has been carefully designed in the construction of the dataset. The collected samples cover a variety of materials (such as pure cotton, spandex, polyester, etc.) and fabric types with a variety of surface density ranges, and images of the front and back sides of each fabric are taken. During the shooting process, the diversity of samples is increased by changing the light intensity, shooting angle, etc. to improve the robustness of the model. A large amount of image data containing fabrics of different types and different organizational structures is collected and labeled. The annotation should include information such as the location of the target area (such as a bounding box), category (such as warp yarn, weft yarn, etc.). Use these data to train a deep learning model.

[0118] Network improvement: An innovative dual-branch network architecture is proposed to process double-sided images of fabrics. This architecture can extract feature information on both sides of the fabric in parallel, overcoming the limitation of traditional methods that rely only on single-sided images for recognition. The front and back sides of a fabric structure are input into the network as a sample for training. By considering the features of both sides of the fabric at the same time, the accuracy and robustness of the tissue structure classification are significantly improved. In order to reduce the consumption of computing resources, the present invention adopts a lightweight convolutional neural network (such as GhostNet, MobileNet, ShuffleNet) as the basic model. While ensuring high performance, this model greatly reduces the number of parameters and computational complexity, making it more suitable for deployment in practical applications, especially in resource-constrained environments. In the feature extraction stage, an attention mechanism module is introduced to correct the weights of features at each level in the channel domain and the spatial domain. This mechanism can enhance the network's ability to capture key features and improve the feature differentiation between different tissues, so that it can still maintain a high classification accuracy under interference such as angle rotation, scale change, and changes in lighting conditions.

[0119] Model training and optimization: Use appropriate optimization algorithms (such as Adam, SGD, etc.) to train the model and continuously adjust the model parameters to improve recognition accuracy and generalization ability. At the same time, use cross-validation and other techniques to evaluate model performance and avoid overfitting.

[0120] Feature extraction and classification: Use the trained deep learning model to perform target detection on the preprocessed fabric image and extract the target area of ​​the weave structure. Then, extract multi-dimensional information such as shape features, texture features, and color features from the target area to form a feature vector. Finally, use a machine learning classifier or a deep learning classifier to classify the feature vector to achieve intelligent recognition of the fabric weave structure.

[0121] It is understandable that the specific implementation of the Web / mobile application interaction module is as follows:

[0122] Front-end design: Design an intuitive and easy-to-use web or mobile user interface. The interface should include image upload area, recognition result display area, operation buttons, etc. Ensure that users can easily upload images, view recognition results, and perform related operations.

[0123] Backend development: Develop a backend server to process user requests, call the API interface of the deep learning recognition module, and manage user data and recognition results. The backend server should have high concurrent processing capabilities to ensure stable operation of the system.

[0124] API interface integration: Integrate the API interface of the deep learning recognition module into the Web / mobile application. After the user uploads the image, the application transmits the image to the backend server, the server calls the API interface for intelligent recognition, and returns the recognition result to the user for display.

[0125] Result display and report generation: Visual display of recognition results is realized in Web / mobile applications, such as marking the detected organizational structure, displaying classification results, etc. At the same time, a report generation function is provided, allowing users to generate test reports according to their needs, including information such as test time, test results, image comparison, etc. The report can support download or print functions, which is convenient for users to archive or share.

[0126] It can be understood that the specific implementation of system integration and optimization is as follows:

[0127] System integration: Integrate the data acquisition module, image processing module, deep learning recognition module and Web / mobile application interaction module to form a complete intelligent detection system for fabric structure, ensuring smooth data transmission, fast processing speed and high accuracy between modules.

[0128] System testing: Conduct comprehensive testing on the integrated system, including functional testing, performance testing, compatibility testing, etc. to ensure that the system can meet user needs and run stably.

[0129] Optimization and adjustment: Continuously optimize the parameters and structure of the deep learning model, introduce new optimization algorithms or regularization techniques to reduce overfitting and improve the generalization ability of the model. At the same time, pay attention to the latest research trends in deep learning, and apply new research results to this system to continuously improve recognition accuracy and efficiency. Optimize the algorithm flow in the image processing module, reduce unnecessary calculation steps, and increase the speed of image processing. At the same time, compress the collected image data to reduce the demand for storage space and transmission bandwidth. Establish a user feedback mechanism to collect user questions and suggestions during use. By regularly analyzing user feedback, timely adjust system functions and interface design to ensure that the system can continue to meet user needs. In order to expand the scope of application of the system, a multi-platform Web / mobile application version can be developed. For example, support multiple platforms such as iOS, Android, and Web browsers at the same time to ensure that users can use this system conveniently on different devices. Strengthen the security design of the system, including data encryption, user identity authentication, access control and other measures. Ensure the security of user data during transmission and storage to prevent data leakage and illegal access. Consider migrating some or all of the system functions to the cloud platform, and use the powerful computing and storage capabilities of cloud services to improve system performance. At the same time, the scalability and flexibility of the system are achieved through cloud services, which is convenient for users to customize and upgrade according to actual needs. Develop automatic update functions to ensure that the system can regularly receive and install the latest software updates and patches. At the same time, provide remote technical support and maintenance services to ensure that the system can operate stably in the long term and adapt to changing market needs. In order to meet the needs of the global market, multi-language support functions can be added to the system. Users can choose the corresponding interface language according to their language preferences to improve the ease of use and internationalization of the system.

[0130] It should be noted that the system can efficiently and accurately identify and classify a variety of textiles, including woven fabrics, weft knitting, warp knitting, knitted fabrics, non-woven fabrics, etc., while providing a user-friendly interactive interface to realize the automation and intelligence of the detection process.

[0131] It should be noted that the application scenarios of this system include but are not limited to:

[0132] Textile production process monitoring: This system is applied to textile production lines to monitor the quality of fabric structure in real time. Once abnormal or unqualified products are found, an alarm will be issued immediately and relevant information will be automatically recorded so that timely measures can be taken to correct them.

[0133] Textile quality inspection: This system is used to inspect the quality of textiles, including key indicators such as fabric density, yarn thickness, and organizational structure. Through fast and accurate test results, it provides textile companies with a powerful means of quality control.

[0134] Textile design and development: In the textile design and development stage, this system is used to quickly test and evaluate design samples. By comparing and analyzing the fabric structure characteristics of different design schemes, valuable reference information is provided to designers.

[0135] Textile education and training: Apply this system to the teaching and training of textile colleges to help students and trainees understand the characteristics and identification methods of fabric structure more intuitively. Improve teaching effect and learning outcomes through practical operation and case analysis.

[0136] Based on this, the fabric structure intelligent detection system and method of the present invention realizes the rapid and accurate identification of various textile structures by integrating portable microscopes, deep learning algorithms and Web / mobile application technologies. The system not only improves the detection efficiency and accuracy, reduces the cost and labor intensity of manual detection, but also provides users with a convenient and efficient detection tool. At the same time, the scalability and customizability of the system also provide strong support for the further intelligent development of the textile industry.

[0137] In addition, if Figure 7 As shown, an embodiment of the present invention also discloses an electronic device, including: at least one processor 210; at least one memory 220, for storing at least one program; when the at least one program is executed by the at least one processor 210, the intelligent detection method of fabric structure in any of the previous embodiments is implemented.

[0138] In addition, an embodiment of the present invention further discloses a computer-readable storage medium, in which computer-executable instructions are stored. The computer-executable instructions are used to execute the intelligent detection method for fabric structure in any of the previous embodiments.

[0139] The system architecture and application scenarios described in the embodiments of the present invention are intended to more clearly illustrate the technical solutions of the embodiments of the present invention, and do not constitute a limitation on the technical solutions provided by the embodiments of the present invention. Those skilled in the art will appreciate that with the evolution of the system architecture and the emergence of new application scenarios, the technical solutions provided by the embodiments of the present invention are equally applicable to similar technical problems.

[0140] Those skilled in the art will appreciate that all or some of the steps in the methods disclosed above, and the functional modules / units in the systems and devices may be implemented as software, firmware, hardware, or a suitable combination thereof.

[0141] In hardware implementation, the division between the functional modules / units mentioned in the above description does not necessarily correspond to the division of physical components; for example, a physical component may have multiple functions, or a function or step may be performed by several physical components in cooperation. Some physical components or all physical components may be implemented as software executed by a processor, such as a central processing unit, a digital signal processor or a microprocessor, or implemented as hardware, or implemented as an integrated circuit, such as an application-specific integrated circuit. Such software may be distributed on a computer-readable medium, which may include a computer storage medium (or non-transitory medium) and a communication medium (or temporary medium). As known to those of ordinary skill in the art, the term computer storage medium includes volatile and non-volatile, removable and non-removable media implemented in any method or technology for storing information (such as computer-readable instructions, data structures, program modules or other data). Computer storage media include, but are not limited to, RAM, ROM, EEPROM, flash memory or other memory technology, CD-ROM, digital versatile disk (DVD) or other optical disk storage, magnetic cassettes, magnetic tapes, disk storage or other magnetic storage devices, or any other medium that can be used to store desired information and can be accessed by a computer. Furthermore, it is well known to those skilled in the art that communication media typically embodies computer readable instructions, data structures, program modules, or other data in a modulated data signal such as a carrier wave or other transport mechanism, and may include any information delivery media.

[0142] The terms "component", "module", "system", etc. used in this specification are used to represent computer-related entities, hardware, firmware, a combination of hardware and software, software, or software in execution. For example, a component can be, but is not limited to, a process running on a processor, a processor, an object, an executable file, an execution thread, a program, or a computer. By way of illustration, both applications running on a computing device and a computing device can be components. One or more components may reside in a process or an execution thread, and a component may be located on a computer or distributed between two or more computers. In addition, these components may be executed from various computer-readable media having various data structures stored thereon. Components may communicate, for example, through local or remote processes based on signals having one or more data packets (e.g., data from two components interacting with another component between a local system, a distributed system, or a network, such as the Internet interacting with other systems through signals).

Claims

1. A method for intelligent detection of fabric structure, characterized in that: Applied to a server, the method comprises: Acquire a fabric image to be detected from a terminal, wherein the terminal is a Web terminal or a mobile terminal, and the fabric image is obtained by taking a photo with a portable microscope; Preprocessing the fabric image to obtain a target fabric image; The target fabric image is input into a pre-trained deep learning model, and the fabric structure result is output, wherein the deep learning model is constructed based on a target detection algorithm, and the deep learning model includes a target detection layer, a feature extraction network and a classifier. The deep learning model outputs the fabric structure result according to the multi-dimensional information of the target fabric image, and the multi-dimensional information includes the weaving rules of the yarns or fibers in the target fabric.

2. The intelligent detection method for fabric structure according to claim 1, characterized in that: The training method of the deep learning model includes: Acquire a data set, wherein the data set includes double-sided images of fabrics of different types and different tissue structures and corresponding weaving rules of yarns or fibers, and the double-sided images of the fabrics include a front side image of the fabric and a back side image of the fabric; Annotating the target area position, category information, and weaving regularity of yarns or fibers in the double-sided fabric image to obtain annotation data; Build a lightweight convolutional neural network; Based on the labeled data, an optimization algorithm is used to perform model training on the lightweight convolutional neural network to adjust target model parameters, wherein the model parameters include weights of features at each level in the channel domain and the spatial domain; The trained deep learning model is obtained according to the target model parameters.

3. The intelligent detection method for fabric structure according to claim 1, characterized in that: The preprocessing of the fabric image to obtain a target fabric image includes: Using Gaussian filtering or median filtering to remove image noise of the fabric image to obtain a first fabric image; Using a contrast enhancement algorithm to adjust the image brightness distribution of the first fabric image to obtain a second fabric image; adjusting the image color of the second fabric image using a color correction algorithm to obtain a third fabric image; Using a threshold-based segmentation algorithm, a region-based segmentation algorithm, or an edge-based segmentation algorithm to segment different tissue regions of the third fabric image to obtain segmented tissue regions; Performing image enhancement processing on the segmented tissue area to obtain a target fabric image, wherein the image enhancement processing includes sharpening filtering and edge detection; Fourier transform and eigenvalue extraction methods are used on the target fabric image to obtain the weaving pattern of the yarns or fibers in the target fabric. The weaving pattern includes the weaving trajectory and direction of the yarns or fibers, which are left-right crossing along the warp, up-and-down crossing along the weft, vertical floating and sinking along the warp, parallel floating and sinking along the weft, and random and disordered multi-directional crossing.

4. The intelligent detection method for fabric structure according to claim 1, characterized in that: The step of inputting the target fabric image into a pre-trained deep learning model and outputting a fabric structure result includes: Using the target detection layer to perform target detection on the target fabric image, and extracting the target area of ​​the fabric structure; The feature extraction network is used to extract multi-dimensional information from the target area, wherein the multi-dimensional information includes shape features, texture features, color features, and weaving rules of yarns or fibers; Generate a feature vector according to the multi-dimensional information; The classifier is used to classify the feature vector to obtain the fabric structure result.

5. The intelligent detection method of fabric structure according to claim 1, characterized in that: The method further comprises: The fabric images are classified and stored according to date and sample number.

6. A method for intelligent detection of fabric structure, characterized in that: Applied to a terminal, the terminal being a Web terminal or a mobile terminal, the method comprises: Acquire a fabric image to be detected, wherein the fabric image is obtained by photographing through a portable microscope; Sending the fabric image to a server so that the server executes the fabric weave structure intelligent detection method according to any one of claims 1 to 5 to generate a fabric weave structure result; The fabric weave structure result is received from the server.

7. A method for intelligently detecting fabric structure, characterized in that: The method further comprises: Display the fabric structure result; A test report is generated according to the fabric structure result and user demand information, and the test report includes test time, test result and image comparison information.

8. A fabric structure intelligent detection system, characterized in that: The system includes a server and a terminal. The terminal and the server interact with each other through an API interface information. The server executes the intelligent detection method for fabric structure as described in any one of claims 1 to 5, and the terminal executes the intelligent detection method for fabric structure as described in any one of claims 6 to 7.

9. An electronic device, comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the intelligent detection method for the fabric structure as described in any one of claims 1 to 5 or the intelligent detection method for the fabric structure as described in any one of claims 6 to 7 is implemented.

10. A computer-readable storage medium storing computer-executable instructions, wherein the computer-executable instructions are used to execute the intelligent detection method for fabric structure according to any one of claims 1 to 5, or the intelligent detection method for fabric structure according to any one of claims 6 to 7.

Images