Sofa fabric defect intelligent detection system and method based on AI image recognition

Abstract

The invention discloses an intelligent sofa fabric defect detection system and method based on AI image recognition, and relates to the technical field of intelligent fabric defect detection.The system comprises a plurality of modules, an image acquisition module is combined with a multispectral camera, quantum enhanced imaging and adaptive optics are introduced, polarization and terahertz cameras are further arranged, the image quality is improved, and the detection accuracy is improved. Image preprocessing adopts deep learning adaptive filtering and the like; aI identification is combined with meta learning, a knowledge graph and data analysis, time sequence analysis is introduced, a VR / AR technology is adopted for result output, a reinforcement learning algorithm is adopted for feedback control, and fabric flaws can be comprehensively and efficiently detected. The system is high in detection precision, can capture flaws in all directions, greatly improves the detection efficiency, meets the requirements of large-scale production, can deeply analyze data, predicts the quality trend, achieves the real-time and accurate production regulation and control, sample management, remote monitoring and other modules, and brings remarkable benefits to enterprises.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent fabric defect detection, and particularly to an intelligent detection system and method for sofa fabric defects based on AI image recognition. Background Art

[0002] In the sofa production industry, fabric quality is one of the key factors determining the quality of sofas. Traditional sofa fabric defect detection mainly relies on manual visual inspection, which has many drawbacks. Manual inspection is inefficient, and the amount of fabric that a skilled worker can inspect in a day is limited, making it difficult to meet the needs of large-scale production. Moreover, long-term visual fatigue easily leads to missed inspections and misjudgments by inspectors, and the detection accuracy cannot be effectively guaranteed, thus affecting the overall quality and brand image of sofas.

[0003] With the development of technology, some enterprises have begun to try to use traditional machine vision technology for fabric defect detection. However, traditional machine vision technology relies on pre-set rules and feature extraction methods, and its adaptability is poor for the complex and variable sofa fabric textures and defect types. For example, when encountering new fabric patterns or subtle defects, traditional machine vision systems often cannot accurately identify them. In addition, traditional machine vision technology has limited capabilities in dealing with interference factors such as light changes and image noise, and is prone to false detections and missed detections, resulting in low reliability of detection results.

[0004] At the same time, existing detection systems lack effective data analysis and feedback control mechanisms. Although they can detect fabric defects, they cannot deeply analyze the detection data and are difficult to discover potential problems existing in the production process. Moreover, after detecting fabric quality problems, they cannot adjust the production equipment in a timely and accurate manner, and cannot achieve real-time optimization of the production process, resulting in low production efficiency and waste of resources. Therefore, developing an efficient, accurate, and intelligent sofa fabric defect detection system has important practical significance. Summary of the Invention

[0005] The intelligent detection system and method for sofa fabric defects based on AI image recognition proposed by the present invention are used to solve the problems mentioned in the above prior art.

[0006] To achieve the above object, the present invention adopts the following technical solution: An intelligent detection system for sofa fabric defects based on AI image recognition, comprising the following modules:

[0007] Image acquisition module: Using a multi-spectral camera combination, covering multiple bands, and collecting fabric images in a Fibonacci spiral layout. The camera exposure time formula is where v is the fabric moving speed, r is the resolution requirement, I is the light intensity, and k is the camera constant; the aperture formula is D is the depth of field, f is the focal length, and c is the imaging constant; the module introduces quantum enhanced imaging technology and an adaptive optical system;

[0008] Image preprocessing module: performs grayscale conversion, filtering and denoising, and sharpening enhancement operations on the image. Grayscale conversion uses the weighted average method, and the grayscale value formula is G = w1R + w2G′ + w3B, where R, G', and B are the pixel values of the red, green, and blue channels respectively; filtering and denoising uses a deep learning adaptive algorithm to retain image details, and sharpening enhancement uses an improved Laplacian operator and combines wavelet transform to analyze the image;

[0009] AI recognition module: constructs a deep convolutional neural network containing a residual network and an attention mechanism, and trains it through sofa fabric defect images. The random gradient descent optimization algorithm is used during the process; meta-learning technology is introduced to adapt to the characteristics of new sofa fabrics. At the same time, prior knowledge of defects is combined with knowledge graph embedding;

[0010] Data analysis module: statistically analyzes defect data, calculates the number of defects n and the area ratio where S i is the area of the i-th defect, and S total is the total area of the fabric. The distribution density uses the analytic hierarchy process to determine the weight of defect types, combines the fuzzy comprehensive evaluation method to evaluate the fabric quality, and introduces the time series analysis method to detect data;

[0011] Result output module: presents the detection results in an intuitive way, viewed through a human-computer interaction interface, supports export in Excel and PDF formats, and uses VR or AR technology to view the results immersively;

[0012] Feedback control module: when the fabric quality grade is lower than the set standard, the system automatically adjusts the device parameters. The rotational speed adjustment amount ΔN = β1×n + β2×S p , where β1 and β2 are coefficients; the reinforcement learning algorithm is introduced to optimize the adjustment strategy.

[0013] Furthermore, it also includes a sample management module, which is used to collect, organize, and annotate sofa fabric defect image samples, establish a sample database, use data augmentation technology to expand the number of samples. At the same time, a generative adversarial network is used to generate virtual defect samples, and the database is updated regularly. An active learning strategy is adopted to select valuable samples for annotation according to the prediction uncertainty of the model.

[0014] Furthermore, it also includes a remote monitoring module, which connects the detection system to a remote server through quantum encryption communication technology to achieve monitoring of the detection results. Managers can view information through a mobile APP or a web page; at the same time, the system supports remote parameter setting and fault diagnosis, and uses edge computing technology to process data on local devices.

[0015] Furthermore, the image acquisition module is also equipped with a polarization camera and a terahertz camera. The polarization camera is used to detect the change in polarization characteristics on the surface of the fabric, identify defects by analyzing the polarization image, and the terahertz camera detects internal defects.

[0016] Furthermore, the AI recognition module adopts federated learning technology to share model parameters among multiple production bases. At the same time, interpretable artificial intelligence technology is introduced to provide explanations for the recognition results of the model. The learning rate is adjusted by where η0 is the initial learning rate, α is the decay coefficient, and t is the number of training iterations.

[0017] Furthermore, the data analysis module combines blockchain technology and Internet of Things technology to store the detection data and quality assessment results on the blockchain and associate them with the Internet of Things data of production equipment. At the same time, it analyzes the impact of each link in the production process on product quality through Internet of Things data analysis.

[0018] Furthermore, a method for an intelligent detection system of sofa fabric defects based on AI image recognition includes the following steps:

[0019] Image acquisition step: Use a multi-spectral camera combination, a polarization camera, and a terahertz camera to acquire images of the sofa fabric. According to the moving speed and surface characteristics of the fabric, adjust the exposure time and aperture size of the camera according to the formulas and and use quantum-enhanced imaging technology and an adaptive optical system to improve the image quality.

[0020] Image preprocessing step: Perform gray conversion, filtering and denoising, and sharpening and enhancement operations on the acquired images. The gray value is calculated according to the formula G = w1R + w2G′ + w3B for gray conversion. Adopt an adaptive filtering algorithm based on deep learning and a method combining an improved Laplacian operator with wavelet transform to highlight the fabric texture and the contour of defects.

[0021] AI recognition step: Input the preprocessed images into a deep convolutional neural network model, and combine residual network, attention mechanism, meta-learning technology, and knowledge graph technology for defect recognition and localization. During the model training process, adjust the learning rate dynamically according to the formula and train with sample images and virtual samples generated by a generative adversarial network to identify defects.

[0022] Data analysis step: Statistically analyze the defect data output by AI recognition, and calculate according to the formulas and Calculate indicators such as the area ratio and distribution density of defects, and use the analytic hierarchy process and fuzzy comprehensive evaluation method to evaluate the overall quality of the sofa fabric. Introduce time series analysis method to predict the changing trend of fabric quality. At the same time, combine blockchain and Internet of Things technologies to associate detection data and production equipment data;

[0023] Result output step: Intuitively present the detection results, including generating fabric images with defect marks, quality assessment reports, and statistical charts. Provide an immersive viewing experience using VR / AR technology. At the same time, support exporting the result data in common formats;

[0024] Feedback control step: According to the detection results and quality assessment levels, perform real-time feedback control on the production process. If the fabric quality level is lower than the set standard, adjust the parameters of the production equipment according to the associated formula, and introduce reinforcement learning algorithms to optimize the adjustment strategy.

[0025] Furthermore, in the image acquisition step, use the characteristics of quantum entanglement to encode the image data collected by the camera. At the same time, through multi-sensor fusion technology, fuse the image data collected by different cameras.

[0026] Furthermore, in the AI recognition step, adopt federated learning technology to collaboratively train models at multiple production bases, and use interpretable artificial intelligence technology to provide visual explanations for the recognition results.

[0027] Furthermore, in the data analysis step, use blockchain smart contract technology. When the detection results trigger the preset quality standards, automatically execute corresponding operations to achieve automated and intelligent management of the production process.

[0028] Compared with the existing technologies, the beneficial effects of the present invention are:

[0029] In terms of detection accuracy, the system uses a combination of multi-spectral cameras, polarization cameras, and terahertz cameras, combined with quantum-enhanced imaging technology and adaptive optical systems, to be able to capture the microscopic and internal defect characteristics of the fabric comprehensively and with high precision. At the same time, the AI recognition module introduces technologies such as meta-learning, knowledge graphs, and interpretable artificial intelligence, greatly improving the recognition accuracy of various complex defects and effectively reducing the situations of missed detection and misjudgment.

[0030] In terms of detection efficiency, the system realizes automated image acquisition and processing, avoiding the low efficiency and fatigue problems of manual detection. Through optimized algorithms and advanced hardware devices, it can quickly process a large amount of image data, greatly improving the detection speed and meeting the needs of large-scale production.

[0031] In terms of data analysis and feedback control, the system utilizes technologies such as time series analysis, blockchain, and the Internet of Things to deeply mine and analyze the detection data, enabling accurate prediction of the changing trends of fabric quality and timely discovery of potential problems in the production process. Meanwhile, the feedback control module introduces a reinforcement learning algorithm, which can adjust the parameters of production equipment in real time and accurately according to the detection results, realizing dynamic optimization of the production process and improving the stability of production efficiency and product quality.

[0032] In addition, the sample management module of the system adopts generative adversarial networks and active learning strategies, enriching sample diversity and improving the adaptability and generalization ability of the model. The remote monitoring module uses quantum encryption communication and edge computing technologies to achieve secure and efficient remote monitoring and management. The virtual reality / augmented reality technology of the result output module provides more intuitive and convenient decision-making support for operators. In short, the patent system brings significant economic and social benefits to sofa fabric production enterprises. Brief Description of the Drawings

[0033] Figure 1 It is a schematic block diagram of the intelligent detection system for sofa fabric defects based on AI image recognition proposed by the present invention;

[0034] Figure 2 It is a bar chart comparing the defect omission rates of different detection methods of the intelligent detection system and method for sofa fabric defects based on AI image recognition proposed by the present invention;

[0035] Figure 3 It is a line chart showing the change of the system detection efficiency over time proposed by the present invention;

[0036] Figure 4 It is a bar chart of the quality prediction accuracy rates of different detection methods of the intelligent detection system and method for sofa fabric defects based on AI image recognition proposed by the present invention. Detailed Embodiments

[0037] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0038] In the description of the present invention, it should be understood that the orientation or positional relationship indicated by terms such as "center", "longitudinal", "transverse", "length", "width", "thickness", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", "clockwise", "counterclockwise", etc. is based on the orientation or positional relationship shown in the drawings. It is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be construed as a limitation to the present invention.

[0039] In addition, the terms "first" and "second" are only used for descriptive purposes and should not be construed as indicating or implying relative importance or implicitly specifying the quantity of the indicated technical features. Thus, the features defined with "first" and "second" may explicitly or implicitly include one or more of the said features. In the description of the present invention, the meaning of "a plurality" is two or more unless otherwise specifically defined. In addition, the terms "mounted", "connected", and "coupled" should be understood in a broad sense. For example, it may be a fixed connection, a detachable connection, or an integral connection; it may be a mechanical connection or an electrical connection; it may be directly connected or indirectly connected through an intermediate medium, and it may be the communication inside two elements. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to specific circumstances. The present invention will be further described in detail below with reference to the drawings.

[0040] Refer to Figures 1-4 : An intelligent detection system for sofa fabric defects based on AI image recognition, including the following modules:

[0041] Image acquisition module: Around the fabric transmission channel, a multi-spectral camera combination is installed according to the Fibonacci spiral layout, covering visible light, infrared, and ultraviolet bands. The camera has automatic focus and optical anti-shake functions, and its exposure time t is dynamically adjusted according to the formula where k is a constant pre-determined according to the camera model and performance. In the camera selected in this embodiment, k = 5×10 7 . The fabric moving speed v is obtained in real time by the transmission device sensor. The image resolution requirement r is set to 15000×12000 pixels according to the detection accuracy, and the ambient light intensity I is measured by the light sensor. The aperture size A is calculated according to the formula Adjustment: The focal length f of the lens can be switched between 50 - 200 mm according to the shooting distance. The depth of field requirement D is set to 5 - 10 cm according to the fabric thickness and detection needs. c is a constant related to the imaging quality, with a value of 2. At the same time, a quantum-enhanced imaging device is introduced. Through a specific quantum entanglement photon pair generation device, entangled photon pairs are generated and used for image acquisition to improve the image resolution and signal-to-noise ratio. An adaptive optical system is equipped. Using a wavefront sensor to monitor the wavefront distortion in the light propagation process in real time, and performing real-time correction through a deformable mirror to ensure that the collected images are clear and distortion-free. In addition, a polarization camera and a terahertz camera are added. The polarization camera is used to capture the changes in the polarization characteristics of the fabric surface, and the terahertz camera can penetrate a certain depth of the fabric to detect internal hidden defects.

[0042] Image preprocessing module: Perform grayscale conversion, filtering denoising, and sharpening enhancement operations on the collected images in sequence. For grayscale conversion, an improved weighted average method is used. Weights are assigned according to the sensitivity of colors to defect features under different spectra. The calculation formula for the grayscale value G is G = w1R + w2G′ + w3B, where R, G', and B are the pixel values of the red, green, and blue channels respectively, and w1, w2, and w3 are the corresponding weights and w1 + w2 + w3 = 1. For filtering denoising, an adaptive filtering algorithm based on deep learning is used. This algorithm learns the characteristics of different types of noise by training a neural network, and can more accurately remove noise while retaining image details. For sharpening enhancement, an improved Laplace operator is used, combined with wavelet transform for multi-scale analysis of the image, highlighting the fabric texture and the contours of defects, and the enhancement effect is more significant.

[0043] AI recognition module: Build a deep convolutional neural network (DCNN) model, combining a residual network (ResNet) and an attention mechanism (Attention Mechanism). The model contains multiple convolutional layers, pooling layers, and fully connected layers, and is trained with a large number of labeled sofa fabric defect images. During the training process, the stochastic gradient descent (SGD) optimization algorithm is used, and the learning rate η is dynamically adjusted. The formula is where η0 is the initial learning rate, α is the decay coefficient, and t is the number of training iterations. Meta-learning technology is introduced to enable the model to quickly adapt to new sofa fabric types and defect characteristics, and can also efficiently learn and recognize when encountering a small number of new samples. At the same time, combined with knowledge graph technology, the prior knowledge of different types of defects is incorporated into the model to improve the accuracy and reliability of recognition. This module can recognize various types of defects, such as holes, stains, color differences, warp and weft misalignments, etc., and accurately determine the location and size of the defects.

[0044] Data analysis module: Statistically analyze the defect data output by the AI recognition module, and calculate indicators such as the number n of defects, the area ratio S p and the distribution density D d etc. The area ratio where \(S\) i is the area of the \(i\)-th defect, and \(S\) total is the total area of the fabric; the distribution density The analytic hierarchy process (AHP) is used to determine the weights of different defect types on the quality of the sofa fabric. Combining the fuzzy comprehensive evaluation method, the overall quality of the sofa fabric is evaluated, and the quality grades are divided into four levels: excellent, good, medium, and poor. The time series analysis method is introduced to analyze the detection data in different time periods, predict the change trend of the fabric quality, and take measures in advance to prevent quality problems.

[0045] Result output module: Develop a dedicated human-computer interaction software interface to present the detection results in an intuitive manner. Through the image drawing algorithm, a fabric image with defect marks is generated, and the defect positions are marked with graphics of different colors. Information such as type and size is displayed in text form beside the image. The quality assessment report is generated in PDF format, including detailed contents such as fabric batch information, detection time, quality grade, the number and proportion of various defects, etc. Statistical charts are in the form of bar charts, line charts, etc., to display information such as the quality trend of different batches of fabrics and the distribution of various defects. The virtual reality (VR) / augmented reality (AR) technology is adopted, using a head-mounted display device and a gesture recognition device, enabling operators to view the detection results in an immersive manner and more intuitively understand the fabric defect situation.

[0046] Feedback control module: According to the detection results and quality assessment grades, perform real-time feedback control on the production process. If the fabric quality grade is lower than the set standard, the system automatically adjusts the parameters of the production equipment, such as the rotational speed \(N\) of the loom, the temperature \(T\) and pressure \(P\) of printing and dyeing, etc. The rotational speed adjustment amount \(\Delta N\) is related to the number of defects \(n\) and the area proportion \(S\) p by the formula \(\Delta N=\beta_1\times n + \beta_2\times S\) p , where \(\beta_1\) and \(\beta_2\) are adjustment coefficients; the temperature adjustment amount \(\Delta T\) and pressure adjustment amount \(\Delta P\) are also calculated according to similar correlation formulas. The reinforcement learning algorithm is introduced to enable the system to automatically optimize the adjustment strategy in the continuous feedback control process, improving the stability of the production process and the product quality.

[0047] In the present invention, a sample management module is also included. This module is used to collect, organize, and annotate the defect image samples of the sofa fabric, and establish a sample database. Data augmentation techniques such as rotation, flipping, and scaling are adopted to expand the number of samples. At the same time, the generative adversarial network (GAN) is used to generate realistic virtual defect samples to further enrich the sample diversity. The sample database is updated and optimized regularly, and the active learning strategy is adopted to select the most valuable samples for annotation according to the prediction uncertainty of the model, so as to improve the accuracy and adaptability of the AI recognition module.

[0048] The present invention also includes a remote monitoring module. The detection system is connected to a remote server through quantum encryption communication technology to achieve real-time and secure monitoring of the detection process and results. Managers can view detection data, quality assessment reports and equipment operating status anytime and anywhere through mobile phone APP or web pages. At the same time, the system supports remote parameter setting and fault diagnosis, and uses edge computing technology to perform partial data processing and analysis on local devices, reducing the amount of data transmission and improving response speed.

[0049] In the present invention, the image acquisition module is also equipped with a polarization camera and a terahertz camera. The polarization camera is used to detect changes in polarization characteristics of the fabric surface. By analyzing the polarization image, some minor defects, such as surface scratches and fluff lodging, can be more accurately identified. The terahertz camera can penetrate a certain depth of the fabric to detect hidden defects inside, such as fiber breakage, interlayer foreign matter, etc., to improve the sensitivity and comprehensiveness of the detection.

[0050] In the present invention, the AI recognition module adopts federated learning technology to share model parameters among multiple production bases, thereby improving the generalization ability of the model while protecting data privacy. At the same time, explainable artificial intelligence technology is introduced to provide a visual explanation for the recognition results of the model, helping operators understand the decision-making process of the model and enhancing trust in the detection results.

[0051] In the present invention, the data analysis module combines blockchain technology and Internet of Things technology to store the test data and quality assessment results on the blockchain and associate them with the Internet of Things data of the production equipment. The decentralized, tamper-proof and traceable characteristics of the blockchain ensure the authenticity and reliability of the test data, and provide strong support for quality traceability and responsibility identification. At the same time, through the analysis of Internet of Things data, we can deeply understand the impact of each link in the production process on product quality and optimize the production process.

[0052] The present invention comprises the following steps:

[0053] Image acquisition steps: Start the multispectral camera combination, polarization camera and terahertz camera, and collect images of the sofa fabric from multiple angles according to the Fibonacci spiral layout. During the fabric transmission process, the fabric movement speed, ambient light intensity and other parameters are obtained in real time. According to the formula and Dynamically adjust the camera's exposure time and aperture size. Use quantum enhanced imaging equipment to generate entangled photon pairs for image acquisition, and enable the adaptive optical system to correct image distortion in real time to ensure that the acquired images are clear and complete and can capture microscopic and internal defect features. At the same time, through the multi-sensor fusion algorithm, the image data collected by different cameras are fused and processed to obtain more comprehensive and accurate fabric information.

[0054] Image preprocessing steps: The collected images are successively subjected to grayscale conversion, filtering denoising, and sharpening enhancement operations. The grayscale value is calculated according to the formula G = w1R + w2G′ + w3B for grayscale conversion. The grayscale image is input into an adaptive filtering neural network model based on deep learning for denoising. Finally, an improved Laplace operator combined with wavelet transform is used to highlight the fabric texture and the contours of defects, improving the image quality and providing high-quality image data for subsequent AI recognition.

[0055] AI recognition steps: The preprocessed images are input into the constructed deep convolutional neural network (DCNN) model. The model combines the residual network (ResNet), attention mechanism (Attention Mechanism), meta-learning technology, and knowledge graph technology for defect recognition and localization. During the training process, the learning rate is dynamically adjusted according to the formula and trained with a large number of labeled sample images and virtual samples generated by the generative adversarial network. At the same time, federated learning technology is adopted to share model parameters among multiple production bases, improving the generalization ability of the model while protecting data privacy. Explainable artificial intelligence technology is used to generate visual explanations for the recognition results of the model to help operators understand the model decision-making process.

[0056] Data analysis steps: For the defect data output by AI recognition, use a data statistics program according to the formula and Calculate indicators such as the area ratio and distribution density of defects. The analytic hierarchy process (AHP) is used to determine the weights of different defect types on the quality of the sofa fabric, and the fuzzy comprehensive evaluation method is combined to evaluate the overall quality of the sofa fabric. The time series analysis method is introduced, and the ARIMA model is used to analyze the detection data in different time periods to predict the change trend of fabric quality. At the same time, through blockchain smart contract technology, when the detection result triggers a preset quality standard, corresponding operations are automatically executed, such as adjusting production parameters, issuing alarms, etc., to realize the automated and intelligent management of the production process.

[0057] Result output steps: Through the human-computer interaction software interface, the detection results are presented in the form of generating fabric images with defect marks, quality assessment reports, and statistical charts. Virtual reality (VR) / augmented reality (AR) technology is used to provide an immersive viewing experience for operators to assist in decision-making. At the same time, the result data is exported in common formats such as Excel and PDF for convenient production management and quality traceability.

[0058] Feedback control step: According to the detection results and quality assessment levels, perform real-time feedback control on the production process. If the fabric quality level is lower than the set standard, adjust the parameters of production equipment, such as loom speed, printing and dyeing temperature and pressure, etc., according to the associated formula. Introduce reinforcement learning algorithms to optimize the adjustment strategy and improve production stability and product quality.

[0059] In the present invention, in the image acquisition step, using the quantum entanglement no-cloning theorem and quantum key distribution technology, perform quantum state encryption coding on multi-dimensional image data, significantly improving the anti-eavesdropping ability and anti-noise interference performance of data transmission. At the same time, adopt a multi-modal sensor fusion architecture, combine multi-spectral and high-resolution camera arrays to collect data in real time, and generate a three-dimensional fabric information model including fiber structure, defect distribution and color parameters through spatio-temporal synchronization algorithms, realizing the intelligent upgrade of textile fabric quality detection.

[0060] In the present invention, in the AI recognition step, adopt a hierarchical federated learning architecture, deploy local model training nodes at each production base, achieve cross-domain knowledge transfer through dynamic weight aggregation strategies, and construct a data security barrier by combining differential privacy and homomorphic encryption technologies, continuously optimizing the model generalization ability while protecting the privacy of production data. Synchronously introduce attention heatmaps and decision tree visualization technologies to construct an end-to-end interpretable AI system, and generate multi-dimensional attribution analysis of detection results through counterfactual explanation technologies, significantly improving the transparency of the quality inspection process and the efficiency of human-machine collaboration.

[0061] In the present invention, in the data analysis step, construct an intelligent contract execution engine based on blockchain, and realize the real-time verification and tamper-proof storage of quality data through distributed ledger technology. When the AI recognition result reaches the preset threshold, the intelligent contract automatically triggers a multi-node consensus mechanism, dynamically adjusts production parameters and issues hierarchical alarms across systems, and at the same time generates audit evidence. This architecture supports seamless docking with Internet of Things devices, realizes self-optimizing management of the production process through automated closed-loop control, significantly improves the response speed and decision-making transparency of the manufacturing system, and ensures full-process traceability.

[0062] The above is only the preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention, according to the technical solution of the present invention and its inventive concept, makes equivalent substitutions or changes, and all should be covered within the protection scope of the present invention.

Claims

1. An intelligent detection system for sofa fabric defects based on AI image recognition, characterized in that, It includes the following modules: Image acquisition module: A multi-spectral camera combination is adopted to cover multiple bands. The fabric image is acquired in a Fibonacci spiral layout. The camera exposure time formula is where v is the fabric moving speed, r is the resolution requirement, I is the light intensity, and k is the camera constant; the aperture formula is D is the depth of field, f is the focal length, and c is the imaging constant; the module introduces quantum enhanced imaging technology and an adaptive optical system; Image preprocessing module: Performs grayscale conversion, filtering and denoising, and sharpening enhancement operations on the image. The grayscale conversion uses the weighted average method, and the grayscale value formula is G = w1R + w2G′ + w3B, where R, G', and B are the pixel values of the red, green, and blue channels respectively; Filtering and denoising uses a deep learning adaptive algorithm to retain image details, and sharpening enhancement uses an improved Laplace operator and combines wavelet transform to analyze the image; AI recognition module: Constructs a deep convolutional neural network with a residual network and an attention mechanism, and trains it through sofa fabric defect images. The stochastic gradient descent optimization algorithm is used during the process; Introduces meta-learning technology to adapt to the characteristics of new sofa fabrics. At the same time, combines knowledge graph embedding with prior knowledge of defects; Data analysis module: Statistically analyze defective data and calculate the number of defects \(n\) and the area ratio where \(S\) i is the area of the \(i\)-th defect, and \(S\) total is the total area of the fabric, and the distribution density Use the analytic hierarchy process to determine the weight of defect types, combine the fuzzy comprehensive evaluation method to evaluate the fabric quality, and introduce the time series analysis method to detect data; Result output module: Presents the detection results in an intuitive way, which can be viewed through a human-computer interaction interface, supports exporting to Excel and PDF formats, and uses VR or AR technology to view the results immersively; Feedback control module: When the fabric quality grade is lower than the set standard, the system automatically adjusts the equipment parameters, and the rotational speed adjustment amount ΔN = β1×n + β2×S p , where β1 and β2 are coefficients; the reinforcement learning algorithm is introduced to optimize the adjustment strategy.

2. The intelligent detection system for defects in sofa fabrics based on AI image recognition according to claim 1, wherein It also includes a sample management module, which is used to collect, organize, and annotate sofa fabric defect image samples, establish a sample database, uses data augmentation technology to expand the number of samples. At the same time, uses a generative adversarial network to generate virtual defect samples, regularly updates the database, and adopts an active learning strategy to select valuable samples for annotation according to the prediction uncertainty of the model.

3. The intelligent detection system for sofa fabric defects based on AI image recognition according to claim 1, characterized in that, It also includes a remote monitoring module, which connects the detection system to a remote server through quantum encryption communication technology to achieve monitoring of the detection results. Managers can view information through a mobile APP or a web page; At the same time, the system supports remote parameter setting and fault diagnosis, and uses edge computing technology to process data on local devices.

4. The intelligent detection system for sofa fabric defects based on AI image recognition according to claim 1, wherein The image acquisition module is also equipped with a polarization camera and a terahertz camera. The polarization camera is used to detect changes in the polarization characteristics of the fabric surface and identify defects by analyzing polarization images. The terahertz camera detects internal defects.

5. The intelligent detection system for defects of sofa fabrics based on AI image recognition according to claim 1, characterized in that, The AI recognition module adopts federated learning technology to share model parameters among multiple production bases. At the same time, interpretable artificial intelligence technology is introduced to provide explanations for the recognition results of the model. The learning rate is adjusted by , where η0 is the initial learning rate, α is the decay coefficient, and t is the number of training iterations.

6. The intelligent detection system for sofa fabric defects based on AI image recognition according to claim 1, characterized in that, The data analysis module combines blockchain technology and Internet of Things technology, stores the detection data and quality assessment results on the blockchain, and associates them with the Internet of Things data of production equipment. At the same time, understands the impact of each link in the production process on product quality through Internet of Things data analysis.

7. A method of applying the intelligent detection system for sofa fabric defects based on AI image recognition according to any one of claims 1-6, characterized in that, It includes the following steps: Image acquisition steps: Use a multi-spectral camera combination, a polarization camera, and a terahertz camera to acquire images of the sofa fabric. According to the moving speed and surface characteristics of the fabric, adjust the exposure time and aperture size of the camera according to the formulas and and use quantum enhanced imaging technology and an adaptive optical system to improve the image quality; Image preprocessing step: Performs grayscale conversion, filtering and denoising, and sharpening enhancement operations on the collected image. The grayscale value is calculated according to the formula G = w1R + w2G′ + w3B. Adopts an adaptive filtering algorithm based on deep learning and a method combining an improved Laplace operator with wavelet transform to highlight the fabric texture and the contour of defects; AI recognition steps: Input the preprocessed image into the deep convolutional neural network model, and combine the residual network, attention mechanism, meta-learning technology and knowledge graph technology to identify and locate defects. During the model training process, adjust the learning rate dynamically according to the formula and train with sample images and virtual samples generated by the generative adversarial network to identify defects. Data analysis steps: Statistically analyze the defective data output by AI recognition, and calculate indicators such as the area proportion and distribution density of defects according to the formulas and Use the analytic hierarchy process and fuzzy comprehensive evaluation method to evaluate the overall quality of the sofa fabric, introduce the time series analysis method to predict the changing trend of the fabric quality. At the same time, combine blockchain and Internet of Things technologies to associate detection data and production equipment data; Result output step: Presents the detection results intuitively, including generating fabric images with defect marks, quality assessment reports, and statistical charts, uses VR / AR technology to provide an immersive viewing experience. At the same time, supports exporting the result data to common formats; Feedback control step: Performs real-time feedback control on the production process according to the detection results and quality assessment levels. If the fabric quality level is lower than the set standard, adjusts the parameters of the production equipment according to the associated formula, and introduces a reinforcement learning algorithm to optimize the adjustment strategy.

8. The intelligent detection method for defects of sofa fabrics based on AI image recognition according to claim 7, characterized in that, In the image acquisition step, the image data collected by the camera is encoded using the characteristics of quantum entanglement. At the same time, through multi-sensor fusion technology, the image data collected by different cameras is fused and processed.

9. The intelligent detection method for defects on sofa fabrics based on AI image recognition according to claim 7, characterized in that, In the AI recognition step, the federated learning technology is used to collaboratively train the model at multiple production bases, and the interpretable artificial intelligence technology is used to provide visual explanations for the recognition results.

10. The intelligent detection method for defects of sofa fabrics based on AI image recognition according to claim 7, characterized in that, In the data analysis step, the blockchain smart contract technology is used. When the detection results trigger the preset quality standards, corresponding operations are automatically executed to achieve the automated and intelligent management of the production process.

Images