Fabric AI quality inspection method and device

Abstract

The invention relates to the technical field of artificial intelligence, and discloses a fabric AI quality inspection method and device, and the method comprises the steps: collecting the image data of a target fabric, carrying out the image enhancement processing of the image data, obtaining the image enhancement data, and extracting the image quality inspection features corresponding to the image enhancement data; performing region division on the image enhancement data to obtain image block regions of the target fabric, and identifying multi-dimensional quality inspection elements corresponding to the image block regions one by one according to image quality inspection features; detecting a multi-dimensional defect element corresponding to the image block area according to the multi-dimensional quality inspection element, and determining a quality inspection factor corresponding to the image block area according to the multi-dimensional defect element; marking a defect area of the target fabric according to the quality inspection factor, and calculating a defect value corresponding to the defect area by using a preset multi-dimensional dynamic quality inspection algorithm; and determining a quality inspection value of the target fabric through the defect value, and generating a quality inspection state of the target fabric according to the quality inspection value. According to the invention, the accuracy of fabric quality inspection can be improved.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to a fabric AI quality inspection method and device. Background Art

[0002] With the advancement of computer vision and deep learning technologies, AI's capabilities in image recognition and processing have been significantly improved, making automated detection of defects in fabrics more accurate and efficient. Different markets and customers may have different standards for fabric quality, and flexible detection methods are needed to adapt to diverse standards for fabric quality inspection.

[0003] Existing fabric quality inspection technologies mostly use high-resolution cameras and image processing software to detect fabric defects, such as color difference and uneven texture. In practical applications, high-resolution cameras are very sensitive to lighting conditions. Under uneven or unstable lighting conditions, image quality may be affected, resulting in inaccurate detection results, which in turn reduces the accuracy of fabric quality inspection. Summary of the invention

[0004] The present invention provides a fabric AI quality inspection method and device, the main purpose of which is to solve the problem of low accuracy in fabric quality inspection.

[0005] To achieve the above object, the present invention provides a fabric AI quality inspection method, comprising:

[0006] Collecting image data of the target fabric, performing image enhancement processing on the image data to obtain image enhancement data, and extracting image quality inspection features corresponding to the image enhancement data;

[0007] Dividing the image enhancement data into regions to obtain image block regions of the target fabric, and identifying multi-dimensional quality inspection elements corresponding to the image block regions one by one according to the image quality inspection features;

[0008] Detecting a multidimensional defect element corresponding to the image block region according to the multidimensional quality inspection element, and determining a quality inspection factor corresponding to the image block region according to the multidimensional defect element;

[0009] Marking the defective area of ​​the target fabric according to the quality inspection factor, and calculating the defect value corresponding to the defective area by using a preset multi-dimensional dynamic quality inspection algorithm, including: counting the number of original defective elements corresponding to the defective area; identifying the number of real-time defective elements corresponding to the defective area according to a preset dynamic time factor; and calculating the defect value corresponding to the defective area according to the number of original defective elements and the number of real-time defective elements by using the following preset multi-dimensional dynamic quality inspection algorithm:

[0010]

[0011] Among them, Q i is the defect value corresponding to the i-th defect area, s i is the severity factor of the i-th defect area, d0 is the number of original defect elements, max is the maximum value function, d it is the number of real-time defective elements in the i-th defective area at the t-th time, and T is the total time;

[0012] The quality inspection value of the target fabric is determined by the defect value, and the quality inspection status of the target fabric is generated according to the quality inspection value.

[0013] Optionally, performing image enhancement processing on the image data to obtain image enhanced data includes:

[0014] Performing denoising processing on the image data to obtain denoised image data;

[0015] Performing contrast adjustment on the denoised image data;

[0016] Performing sharpening processing on the denoised image data after contrast adjustment;

[0017] The denoised image data after sharpening is subjected to color correction to obtain image enhancement data.

[0018] Optionally, extracting image quality inspection features corresponding to the image enhancement data includes:

[0019] Generating a color histogram corresponding to the image enhancement data;

[0020] Calculating the color distribution of the image enhancement data according to the color histogram;

[0021] Extracting color quality inspection features corresponding to the image enhancement data according to the color distribution;

[0022] Identifying crack features corresponding to the image enhancement data;

[0023] Extracting damage quality inspection features corresponding to the image enhancement data according to the crack features;

[0024] The color quality inspection feature and the damage quality inspection feature are merged into an image quality inspection feature.

[0025] Optionally, dividing the image enhancement data into regions to obtain image block regions of the target fabric includes:

[0026] Counting the total number of pixels corresponding to the image enhancement data;

[0027] Generate a division unit corresponding to the image enhancement data according to the total number of pixels;

[0028] The image enhancement data is divided into regions according to the division unit to obtain image block regions of the target fabric.

[0029] Optionally, the identifying the multi-dimensional quality inspection elements corresponding to the image block regions one by one according to the image quality inspection features includes:

[0030] Identify the color quality inspection sub-features and damage quality inspection sub-features corresponding to the image block areas one by one;

[0031] Comparing the color quality inspection feature in the image quality inspection feature with the color quality inspection sub-feature to obtain a quality inspection color element and a quality inspection stain element;

[0032] Comparing the damage quality inspection feature in the image quality inspection feature with the damage quality inspection sub-feature to obtain a quality inspection damage element and a quality inspection texture element;

[0033] The quality inspection color element, the quality inspection stain element, the quality inspection damage element and the quality inspection texture element are merged into a multi-dimensional quality inspection element corresponding to the image block area.

[0034] Optionally, detecting the multidimensional defect element corresponding to the image block area according to the multidimensional quality inspection element includes:

[0035] Obtain target multi-dimensional quality inspection elements of target quality inspection fabric;

[0036] Comparing the multidimensional quality inspection elements with the target multidimensional quality inspection elements one by one to obtain comparison factors;

[0037] The multi-dimensional defect factor corresponding to the image block area is determined according to the contrast factor.

[0038] Optionally, determining the quality inspection factor corresponding to the image block area according to the multi-dimensional defect element includes:

[0039] Counting the number of defect elements corresponding to the multi-dimensional defect elements corresponding to the image block area;

[0040] When the number of defective elements is greater than a preset element number threshold, determining the quality inspection factor corresponding to the image block area as a preset first quality inspection factor;

[0041] When the number of defective elements is equal to a preset element number threshold, the quality inspection factor corresponding to the image block area is determined as a preset second quality inspection factor.

[0042] Optionally, marking the defective area of ​​the target fabric according to the quality inspection factor includes:

[0043] When the quality inspection factor is the first quality inspection factor, marking the image block area corresponding to the quality inspection factor as a defective area;

[0044] When the quality inspection factor is the second quality inspection factor, the image block area corresponding to the quality inspection factor is marked as a non-defective area.

[0045] Optionally, determining the quality inspection value of the target fabric according to the defect value includes:

[0046] Determining the total area quantity according to the defective area and the non-defective area of ​​the target fabric;

[0047] Determining the quality inspection feedback factor of the target fabric through a preset subjective requirement;

[0048] The quality inspection value of the target fabric is calculated by the defect value, the total area number and the quality inspection feedback factor:

[0049]

[0050] Wherein, Z is the quality inspection value, δ j is the quality inspection feedback factor of the target fabric corresponding to the jth subjective user, Q i is the defect value corresponding to the i-th defect area, n is the number of defect areas, H is the total number of areas, γ is the number of element quality inspections of the target fabric, and m is the number of subjective users.

[0051] In order to solve the above problems, the present invention also provides a fabric AI quality inspection device, the device comprising:

[0052] An image quality inspection feature extraction module is used to collect image data of the target fabric, perform image enhancement processing on the image data to obtain image enhancement data, and extract image quality inspection features corresponding to the image enhancement data;

[0053] A multi-dimensional quality inspection element identification module is used to divide the image enhancement data into regions to obtain image block regions of the target fabric, and identify multi-dimensional quality inspection elements corresponding to the image block regions one by one according to the image quality inspection features;

[0054] A quality inspection factor determination module, configured to detect a multidimensional defect element corresponding to the image block region according to the multidimensional quality inspection element, and determine a quality inspection factor corresponding to the image block region according to the multidimensional defect element;

[0055] A defect value calculation module, used for marking the defect area of ​​the target fabric according to the quality inspection factor, and calculating the defect value corresponding to the defect area by using a preset multi-dimensional dynamic quality inspection algorithm;

[0056] The quality inspection status generating module is used to determine the quality inspection value of the target fabric according to the defect value, and to generate the quality inspection status of the target fabric according to the quality inspection value.

[0057] The embodiments of the present invention obtain a detailed view of the target fabric to provide basic data for subsequent processing, improve image quality, enhance details, and make defects easier to identify; obtain useful information and features to help accurately evaluate the quality of the fabric, divide the image into small blocks to make the detection more precise and facilitate the discovery of local defects, analyze the characteristics of each area, and provide multi-dimensional quality information; convert defect data into quality inspection factors to help quantify the overall quality of the fabric, quantify the severity of the defects, and provide an objective basis for quality evaluation: combine all data to give a final quality assessment to help make decisions. Therefore, the fabric AI quality inspection method and device proposed in the present invention can solve the problem of low accuracy when performing fabric quality inspection. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] Figure 1 A schematic diagram of a fabric AI quality inspection method according to an embodiment of the present invention;

[0059] Figure 2 A schematic diagram of a process of area division provided by an embodiment of the present invention;

[0060] Figure 3 A schematic diagram of a process for detecting multi-dimensional defect elements provided by an embodiment of the present invention;

[0061] Figure 4 A functional module diagram of a fabric AI quality inspection device provided by an embodiment of the present invention;

[0062] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0063] 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.

[0064] The embodiment of the present application provides a fabric AI quality inspection method. The execution subject of the fabric AI quality inspection method includes but is not limited to at least one of the electronic devices such as a server, a terminal, etc. that can be configured to execute the method provided in the embodiment of the present application. In other words, the fabric AI quality inspection method can be executed by software or hardware installed on a terminal device or a server device, and the software can be a blockchain platform. The server includes but is not limited to: a single server, a server cluster, a cloud server or a cloud server cluster, etc. The server can be an independent server, or it can be a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, content delivery networks (Content Delivery Network, CDN), and big data and artificial intelligence platforms.

[0065] Reference Figure 1 FIG. 1 is a flow chart of a fabric AI quality inspection method provided by an embodiment of the present invention. In this embodiment, the fabric AI quality inspection method includes:

[0066] S1. Collect image data of a target fabric, perform image enhancement processing on the image data to obtain image enhancement data, and extract image quality inspection features corresponding to the image enhancement data.

[0067] In an embodiment of the present invention, the image data refers to a fabric surface image obtained by photographing or scanning technology, which usually contains fabric details such as texture, color, structure, etc., and is used to identify defects and evaluate the quality of the fabric. The image data of the target fabric can be obtained from a pre-stored storage area through a computer statement with a data capture function (such as a Java statement, a Python statement, etc.).

[0068] Furthermore, in order to improve the visual effect of the image data of the target fabric or enhance specific information in the image, it is necessary to perform enhancement processing on the image data of the target fabric, so as to make the details in the image clearer and improve the visibility of important features in the image data.

[0069] In the embodiment of the present invention, the image enhancement data refers to the image obtained by processing the collected original image data of the target fabric through denoising, contrast adjustment, brightness enhancement, sharpening, etc., so as to improve the image quality and detail clarity, so as to perform subsequent analysis and defect detection more accurately.

[0070] In the embodiment of the present invention, performing image enhancement processing on the image data to obtain image enhanced data includes:

[0071] Performing denoising processing on the image data to obtain denoised image data;

[0072] Performing contrast adjustment on the denoised image data;

[0073] Performing sharpening processing on the denoised image data after contrast adjustment;

[0074] The denoised image data after sharpening is subjected to color correction to obtain image enhancement data.

[0075] In detail, the image data is denoised, that is, the noise in the image is removed to improve the clarity of the image. Random noise can be removed by Gaussian filtering to obtain denoised image data. Based on the denoised image data, the difference in brightness values ​​in the image is enhanced to make the contrast between different regions more obvious. The grayscale distribution of the image can be adjusted to enhance the image contrast. The denoised image data after contrast adjustment has more obvious regional differences, which helps to highlight the image details. The image is then sharpened to improve the edge clarity of the image and make the details more prominent. The sharpening filter can be used for convolution processing to enhance the high-frequency part of the image and improve the edge clarity. Finally, the color of the image is adjusted to ensure the accuracy and naturalness of the color. The color deviation caused by different light sources in the image can be corrected to make the color of the color-corrected image data more natural and accurate, thereby achieving the final image enhancement effect.

[0076] Furthermore, based on the image data after image enhancement processing, the features in the image can be analyzed to identify and locate defects in the image corresponding to the target fabric, such as color distortion, cracks, etc., so as to analyze the overall quality of the target fabric.

[0077] In the embodiment of the present invention, the image quality inspection features are key indicators used to analyze image quality to determine whether the image meets specific quality standards and requirements, including color features and damage features.

[0078] In an embodiment of the present invention, extracting the image quality inspection features corresponding to the image enhancement data includes:

[0079] Generating a color histogram corresponding to the image enhancement data;

[0080] Calculating the color distribution of the image enhancement data according to the color histogram;

[0081] Extracting color quality inspection features corresponding to the image enhancement data according to the color distribution;

[0082] Identifying crack features corresponding to the image enhancement data;

[0083] Extracting damage quality inspection features corresponding to the image enhancement data according to the crack features;

[0084] The color quality inspection feature and the damage quality inspection feature are merged into an image quality inspection feature.

[0085] In detail, the color histogram of the image enhancement data is calculated to obtain the pixel distribution of each color, such as generating a histogram through an image processing library, and then analyzing the color distribution in the image according to the color histogram, counting the frequency of occurrence of each color, and calculating the statistics of the color distribution (such as mean, variance), and extracting color-related quality inspection features from the color distribution, such as color uniformity, saturation, etc.

[0086] In an embodiment of the present invention, extracting the color quality inspection features corresponding to the image enhancement data according to the color distribution includes:

[0087] Extracting image color difference data in the color distribution;

[0088] Determine the stain features and color difference features in the image enhancement data according to the image color difference data;

[0089] A color quality inspection feature corresponding to the image enhancement data is determined according to the stain feature and the color difference feature.

[0090] In detail, the color value of each pixel in the image (such as RGB or Lab color space) is calculated, and the color distribution is analyzed using statistical methods or color models (such as histogram equalization). Abnormal points in the color distribution that obviously deviate from the main color pattern are identified. These points can be marked as stains, and the difference between the color values ​​is calculated. For example, a color difference formula (such as ΔE) is used to evaluate the color change, so as to obtain the color quality inspection features corresponding to the image enhancement data, and then the stain features in the image can be determined according to the image color difference in the color distribution. The color quality inspection features include color difference features and stain features.

[0091] Specifically, cracks in images are identified based on a crack detection algorithm, that is, image processing techniques such as boundary detection and morphological operations are used to extract crack information, and damage-related quality inspection features such as crack length, width and density are calculated based on crack features to evaluate the severity of the damaged area in the image. The weaving features of the fabric can be determined based on the crack features, that is, for fabrics with cracks, the weaving features of the fabric are non-uniform, and for fabrics without cracks, the weaving features of the fabric are uniform, thereby obtaining damage quality inspection features corresponding to the image enhancement data, and the damage quality inspection features include crack features and weaving features, and then the color quality inspection features and the damage quality inspection features are fused, and the two quality inspection information are comprehensively considered through feature splicing to fuse them into the final image quality inspection features.

[0092] Furthermore, in order to be able to perform detailed analysis of various areas of the target fabric and identify and deal with local problems, the large image needs to be divided into small blocks, so that quality problems of the fabric, such as defects or color unevenness, can be detected more accurately.

[0093] S2. Divide the image enhancement data into regions to obtain image block regions of the target fabric, and identify multi-dimensional quality inspection elements corresponding to the image block regions one by one according to the image quality inspection features.

[0094] In the embodiment of the present invention, the image block area refers to dividing the entire image into multiple small areas or blocks, each area represents a part of the fabric, allowing different areas of the fabric to be processed and analyzed independently, thereby improving the accuracy of image enhancement and quality detection.

[0095] In the embodiment of the present invention, refer to Figure 2 As shown, the image enhancement data is divided into regions to obtain the image block region of the target fabric, including:

[0096] S21, counting the total number of pixels corresponding to the image enhancement data;

[0097] S22, generating a division unit corresponding to the image enhancement data according to the total number of pixels;

[0098] S23, performing region division on the image enhancement data according to the division unit to obtain image block regions of the target fabric.

[0099] In detail, the total number of pixels of an image can be calculated by the width and height of the image. For example, if the width of the image is W and the height is H, the total number of pixels is W*H. The size of each division unit is determined according to the total number of pixels and the required area size. For example, if the image is divided into N areas, the width and height of each area can be selected appropriately to ensure that they cover the entire image. The goal is to divide the image into N units, and the unit width is W / sqrt(N), and the unit height is H / sqrt(N). Thus, the image block area after area division can be obtained. The image can be divided into several rows and columns, and the size of each unit is unit width x unit height.

[0100] For example, if there is an image with a size of 800x600 pixels, the total number of pixels = width x height = 800x 600 = 480,000 pixels, and the image is divided into 16 uniform small blocks, then the number of division units = 16, the width of each unit = width / √number of division units = 800 / 4 = 200 pixels, the height of each unit = height / √number of division units = 600 / 4 = 150 pixels, then according to the calculated unit size, the image is divided into small blocks of 200x150 pixels.

[0101] Furthermore, in order to more accurately identify problem areas in images and improve the accuracy of quality inspection, it is necessary to refine and identify features such as color and damage. This helps to fully understand the image quality, including color consistency, damage, stains and texture status, so as to obtain complete quality information.

[0102] In the embodiment of the present invention, the multi-dimensional quality inspection elements refer to different types of quality inspection feature information corresponding to each image block area, including color features, damage features, stain features and texture features. These elements are combined to provide a comprehensive quality assessment of the image area.

[0103] In the embodiment of the present invention, the step of identifying the multi-dimensional quality inspection elements corresponding to the image block regions one by one according to the image quality inspection features includes:

[0104] Identify the color quality inspection sub-features and damage quality inspection sub-features corresponding to the image block areas one by one;

[0105] Comparing the color quality inspection feature in the image quality inspection feature with the color quality inspection sub-feature to obtain a quality inspection color element and a quality inspection stain element;

[0106] Comparing the damage quality inspection feature in the image quality inspection feature with the damage quality inspection sub-feature to obtain a quality inspection damage element and a quality inspection texture element;

[0107] The quality inspection color element, the quality inspection stain element, the quality inspection damage element and the quality inspection texture element are merged into a multi-dimensional quality inspection element corresponding to the image block area.

[0108] In detail, each image block area is analyzed to identify the color-related quality inspection sub-features and the damage-related quality inspection sub-features, wherein the color quality inspection sub-features include color features and stain features, and the damage quality inspection sub-features include texture features and damage features. The color quality inspection features in the image (such as color uniformity, color saturation, etc.) are compared with the identified color quality inspection sub-features. Through this comparison, specific quality inspection color elements (such as color unevenness, color deviation, etc.) and quality inspection stain elements (such as stains, pollution, etc.) can be obtained, and the damage quality inspection features in the image (such as damage) can be compared with the identified color quality inspection sub-features. The damaged quality inspection sub-features (such as cracks, tearing degree, etc.) are compared with the identified damaged quality inspection sub-features to obtain quality inspection damage elements (such as cracks, tearing degree, etc.) and quality inspection texture elements (such as texture loss, deformation, etc.), and the obtained quality inspection color elements, quality inspection stain elements, quality inspection damage elements and quality inspection texture elements are combined to form a multi-dimensional quality inspection element of the image block area. The fusion process provides comprehensive quality inspection information on the image block area, which is helpful to evaluate and improve fabric quality, and can comprehensively identify and analyze various defects and problems in fabric images, thereby improving the accuracy and effect of quality inspection.

[0109] Furthermore, different quality inspection elements cover multiple dimensions of fabric quality. Multidimensional quality inspection elements provide information on multiple aspects including color, damage, stains and texture. By detecting these elements, we can fully understand the quality of the image block area and identify potential defects. Based on multidimensional quality inspection elements, we can quantify various defect factors, such as the degree of color deviation, the area of ​​damage, etc., thereby providing more accurate quality assessment and more precisely identifying and locating quality problems.

[0110] S3. Detect a multidimensional defect element corresponding to the image block region according to the multidimensional quality inspection element, and determine a quality inspection factor corresponding to the image block region according to the multidimensional defect element.

[0111] In an embodiment of the present invention, the multidimensional defect element refers to multiple different dimensions or features used to describe and identify defects during the quality inspection process, such as color, texture, morphology, and stains, wherein color includes color difference, the difference between the actual color and the standard or expected color, color spots, and color abnormalities in local areas, such as stains or uneven color; texture includes inconsistent weaves, irregularities in the texture of the fabric surface, such as broken lines or missing textures, and fabric structural defects, such as brushing and holes in the fabric; morphology includes damage, such as physical damage such as tears, cracks or wear, and deformation, and the shape of the material deviates from expectations, such as wrinkles or deformations; stains include stains, stains or contaminants on the material, such as oil stains, mud stains, etc., and foreign matter, foreign matter embedded in the material, such as dust or fiber fragments.

[0112] In the embodiment of the present invention, refer to Figure 3 As shown, the detecting the multi-dimensional defect element corresponding to the image block area according to the multi-dimensional quality inspection element includes:

[0113] S31, obtaining target multi-dimensional quality inspection elements of the target quality inspection fabric;

[0114] S32, comparing the multidimensional quality inspection elements with the target multidimensional quality inspection elements one by one to obtain comparison factors;

[0115] S33: Determine a multi-dimensional defect factor corresponding to the image block area according to the comparison factor.

[0116] In detail, for the target quality inspection fabric (such as standard or reference fabric), its multi-dimensional quality inspection elements are obtained, which include color, texture, morphology, stains and other characteristics, as a quality benchmark for comparison, then a colorimeter or color sensor is used to obtain the color data of the reference fabric, determine the standard color value (such as RGB value, LAB value, etc.), record the color distribution on the fabric, including any color difference or gradient, and record the characteristics and location of various types of damage (tears, cracks, wear, etc.) that may exist on the fabric, measure and record the size, shape and location of the damage; identify different types of stains (such as oil stains, mud stains, etc.) and their characteristics, and record the distribution area, size and shape of the stains; record the standard weave pattern of the fabric, including the weaving structure and texture direction of the fabric, and use corresponding detection equipment and technology (such as image processing, color analysis, microscopy, etc.) to obtain the quality inspection element data of the sample.

[0117] Specifically, the quality inspection elements (such as color features, damage features, etc.) of the image block area are compared with the target multidimensional quality inspection elements of the target quality inspection fabric one by one. The comparison process identifies the differences or deviations between the image block area and the target quality inspection fabric, and then determines the multidimensional defect factors corresponding to the image block area based on the comparison results. The defect factors may include color deviation, degree of damage, stain location or weave abnormality, etc., thereby helping to identify and quantify specific quality problems in the image block area.

[0118] Furthermore, by carefully analyzing the defective elements in the image block area, the quality of the fabric or product can be more accurately evaluated, and by counting and analyzing the number of defective elements, the possible quality problems of the fabric can be better understood, thereby identifying specific areas of the fabric that need improvement.

[0119] In the embodiment of the present invention, the quality inspection factor refers to an indicator used to evaluate the quality of a product or material during the quality inspection process of the fabric, which is helpful for identifying whether the fabric product meets the standards and specifications.

[0120] In the embodiment of the present invention, determining the quality inspection factor corresponding to the image block area according to the multi-dimensional defect element includes:

[0121] Counting the number of defect elements corresponding to the multi-dimensional defect elements corresponding to the image block area;

[0122] When the number of defective elements is greater than a preset element number threshold, determining the quality inspection factor corresponding to the image block area as a preset first quality inspection factor;

[0123] When the number of defective elements is equal to a preset element number threshold, the quality inspection factor corresponding to the image block area is determined as a preset second quality inspection factor.

[0124] In detail, the image block area is detected, the number of defective elements is counted, and the number of defective elements in each image block area is recorded. For example, if the defective elements corresponding to image block area 1 are color, texture, and morphology, the number of defective elements is 3. If the defective elements corresponding to image block area 2 are color, texture, morphology, and stains, the number of defective elements is 4. When the number of defective elements is greater than zero, the quality inspection factor of the image block area is set to 1. When the number of defective elements is equal to zero, the quality inspection factor of the image block area is set to 0. The element number threshold is 0, the first quality inspection factor is a value of 1, and the second quality inspection factor is a value of 0. The quality inspection factor of each image block area can be effectively associated with the number of its defective elements.

[0125] Furthermore, by corresponding different quality inspection factors to specific defective or non-defective areas, defects in the fabric can be more accurately identified and marked, thereby reducing errors in subjective judgment and making defect identification more objective and consistent. After determining the defective area based on the quality inspection factor, it is not necessary to check the entire fabric piece by piece, thereby improving efficiency.

[0126] S4. Mark the defective area of ​​the target fabric according to the quality inspection factor, and calculate the defect value corresponding to the defective area using a preset multi-dimensional dynamic quality inspection algorithm.

[0127] In an embodiment of the present invention, the defective area refers to a part identified on a product or fabric that does not meet quality standards or has defects, such as cracks, stains, color differences, weaving defects, etc. The defective area is the focus of marking and processing during the quality inspection process, and affects the overall quality and performance of the product.

[0128] In an embodiment of the present invention, marking the defective area of ​​the target fabric according to the quality inspection factor includes:

[0129] When the quality inspection factor is the first quality inspection factor, marking the image block area corresponding to the quality inspection factor as a defective area;

[0130] When the quality inspection factor is the second quality inspection factor, the image block area corresponding to the quality inspection factor is marked as a non-defective area.

[0131] In detail, if the quality inspection factor is the first quality inspection factor, that is, the quality inspection factor is a value of 1, the defect type or defect index on the fabric, then the corresponding image block area will be marked as a defect area, indicating that there is a problem in the area and needs attention or repair; if the quality inspection factor is the second quality inspection factor, that is, the quality inspection factor is a value of 0, the normal part of the fabric or the defect-free standard, then the corresponding image block area will be marked as a non-defective area, indicating that the area meets the quality standards and does not require further processing. Therefore, the type of quality inspection factor determines the marking status of the fabric image block, so as to accurately identify and distinguish defective and non-defective areas.

[0132] Furthermore, by combining the original defects and real-time data, the severity and development trend of the defects can be comprehensively evaluated, the changes in defects can be tracked in real time, and quality problems in the production process can be discovered and responded to in a timely manner. The defect values ​​corresponding to the defective areas in the fabric can be calculated based on real-time detection to identify the problem areas and optimize the fabric production process and quality inspection control measures.

[0133] In the embodiment of the present invention, the defect value refers to an indicator used to quantify and evaluate the quality problem of a specific defect area, and is calculated based on multiple factors to reflect the severity and impact of the defect.

[0134] In the embodiment of the present invention, the method of calculating the defect value corresponding to the defect area by using a preset multi-dimensional dynamic quality inspection algorithm includes:

[0135] Counting the number of original defect elements corresponding to the defect area;

[0136] Identifying the number of real-time defective elements corresponding to the defective area according to a preset dynamic time factor;

[0137] The defect value corresponding to the defect area is calculated according to the original defect element quantity and the real-time defect element quantity using the following preset multi-dimensional dynamic quality inspection algorithm:

[0138]

[0139] Among them, Q i is the defect value corresponding to the i-th defect area, s i is the severity factor of the i-th defect area, d0 is the number of original defect elements, max is the maximum value function, d it is the real-time number of defective elements in the i-th defective area at the t-th time, and T is the total time.

[0140] Specifically, the number of defective elements d0 initially detected in the defective area is recorded, and the real-time number of defective elements d at each time point t is recorded according to the dynamic time factor. it , the quality inspection features of the target fabric are continuously extracted according to the changes in time points, and then the real-time defective area of ​​the target fabric is determined according to the quality inspection features, and the number of defective elements corresponding to the defective area is determined, that is, the step of obtaining the number of defective elements, that is, the real-time number of defective elements, is repeated at each moment, and the real-time defect value is calculated according to the multi-dimensional dynamic quality inspection algorithm.

[0141] Specifically, the severity factor s of the i-th defect area is i Indicates when s i = 0 means no defect, when s i=1 indicates the maximum defect, then first determine the severity factor corresponding to the defect area. When the area initially indicates no defect, then s i =0, then Q i = 0, and when the region is initially a defective region, then s i =1, and select the maximum number of defect elements identified in the real-time defect value, and select the maximum value between the real-time defect element number and the initial element number to obtain the defect value corresponding to the defect area. For example, when the defect area is initially d0=3, s i =1, and because of the light problem in the initial detection, the number of defective elements detected at the initial time may be inaccurate, so the maximum number of defective elements detected in real time is selected. If the maximum number of defective elements detected in real time is 4, the defect value corresponding to the final defective area is 4. If the area measured at the beginning is a defective area, and no defective elements are detected in the preset detection number threshold in the subsequent detection process, then s i = 0, otherwise s i =1, when s i =1, the defect value of the area is calculated based on the initial defect element number and the real-time defect element number.

[0142] Furthermore, by determining the quality inspection value of the target fabric through the defect value, the actual defect situation of the fabric and the subjective evaluation can be considered comprehensively, so as to obtain a comprehensive quality assessment, which combines the objective defect data and subjective user feedback to make the quality inspection result more comprehensive and accurate.

[0143] S5. Determine the quality inspection value of the target fabric according to the defect value, and generate the quality inspection status of the target fabric according to the quality inspection value.

[0144] In the embodiment of the present invention, the quality inspection value refers to a numerical value for quantitatively evaluating the quality of a product or material, and is intended to provide a comprehensive quality standard based on multiple factors such as product defects, performance indicators, user feedback, etc.

[0145] In an embodiment of the present invention, determining the quality inspection value of the target fabric according to the defect value includes:

[0146] Determining the total area quantity according to the defective area and the non-defective area of ​​the target fabric;

[0147] Determining the quality inspection feedback factor of the target fabric through a preset subjective requirement;

[0148] The quality inspection value of the target fabric is calculated by the defect value, the total area number and the quality inspection feedback factor:

[0149]

[0150] Wherein, Z is the quality inspection value, δ j is the quality inspection feedback factor of the target fabric corresponding to the jth subjective user, Q i is the defect value corresponding to the i-th defect area, n is the number of defect areas, H is the total number of areas, γ is the number of element quality inspections of the target fabric, and m is the number of subjective users.

[0151] Specifically, the total number of defective areas and non-defective areas in the target fabric is counted, and the total quality inspection quantity is determined according to the total number of areas and the element quality inspection quantity of the target fabric. For example, if the element quality inspection quantity is 4 and the total number of areas is 6, the total quality inspection quantity is 24. Then, the quality inspection feedback factor is set according to the needs of each subjective user. The quality inspection feedback factor is used to adjust the quality inspection value to meet the subjective evaluation standards of different users. Calculate the average value of the subjective user feedback factor, that is, each user gives the target fabric a quality inspection score, then the quality inspection feedback factor is the quality inspection score of each user, and then take the average of the quality inspection scores corresponding to all the users who scored, and calculate the defect value Q corresponding to the defect area i The quality inspection value of the target fabric is determined based on the average value and the proportion of the total quality inspection quantity H×γ. The final quality inspection value Z combines subjective feedback and actual defect data to comprehensively evaluate the overall quality of the fabric.

[0152] Specifically, the quality inspection status of the target fabric is generated according to the quality inspection value, and the quality inspection status includes excellent, the quality inspection value is higher than the expected standard and the fabric quality is excellent; good, the quality inspection value is within the standard range and the fabric quality meets the requirements; medium, the quality inspection value is close to or slightly lower than the standard range, the quality is medium and needs to be improved; poor, the quality inspection value is lower than the standard range, the fabric quality does not meet the requirements and needs to be re-inspected or processed, that is, the quality inspection value is directly compared with the set standard value to obtain the quality inspection status of the target fabric. For example, the calculated quality inspection value of the fabric is 75, and the preset standard range is 70-80. According to this standard, the quality inspection status of the fabric can be evaluated as "good". If the quality inspection value is lower than 70, it may be evaluated as "medium" or "poor", and measures need to be taken to improve it.

[0153] The embodiments of the present invention obtain a detailed view of the target fabric to provide basic data for subsequent processing, improve image quality, enhance details, and make defects easier to identify; obtain useful information and features to help accurately evaluate the quality of the fabric, divide the image into small blocks to make the detection more precise and facilitate the discovery of local defects, analyze the characteristics of each area, and provide multi-dimensional quality information; convert defect data into quality inspection factors to help quantify the overall quality of the fabric, quantify the severity of the defects, and provide an objective basis for quality evaluation: combine all data to give a final quality assessment to help make decisions. Therefore, the fabric AI quality inspection method and device proposed in the present invention can solve the problem of low accuracy when performing fabric quality inspection.

[0154] like Figure 4 , which is a functional module diagram of a fabric AI quality inspection device provided by an embodiment of the present invention.

[0155] The fabric AI quality inspection device 100 of the present invention can be installed in an electronic device. According to the functions to be implemented, the fabric AI quality inspection device 100 can include an image quality inspection feature extraction module 101, a multi-dimensional quality inspection element recognition module 102, a quality inspection factor determination module 103, a defect value calculation module 104 and a quality inspection status generation module 105. The module of the present invention can also be called a unit, which refers to a series of computer program segments that can be executed by an electronic device processor and can complete fixed functions, which are stored in the memory of the electronic device.

[0156] In this embodiment, the functions of each module / unit are as follows:

[0157] The image quality inspection feature extraction module 101 is used to collect image data of the target fabric, perform image enhancement processing on the image data to obtain image enhancement data, and extract image quality inspection features corresponding to the image enhancement data;

[0158] The multi-dimensional quality inspection element identification module 102 is used to divide the image enhancement data into regions to obtain image block regions of the target fabric, and identify the multi-dimensional quality inspection elements corresponding to the image block regions one by one according to the image quality inspection features;

[0159] The quality inspection factor determination module 103 is used to detect the multidimensional defect element corresponding to the image block area according to the multidimensional quality inspection element, and determine the quality inspection factor corresponding to the image block area according to the multidimensional defect element;

[0160] The defect value calculation module 104 is used to mark the defect area of ​​the target fabric according to the quality inspection factor, and calculate the defect value corresponding to the defect area using a preset multi-dimensional dynamic quality inspection algorithm;

[0161] The quality inspection status generating module 105 is used to determine the quality inspection value of the target fabric according to the defect value, and generate the quality inspection status of the target fabric according to the quality inspection value.

[0162] In detail, each module described in the fabric AI quality inspection device 100 in the embodiment of the present invention is used in the same manner as described above. Figures 1 to 3 The fabric AI quality inspection method described in the invention has the same technical means and can produce the same technical effects, so I will not go into details here.

[0163] In the several embodiments provided by the present invention, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are only illustrative, for example, the division of the modules is only a logical function division, and there may be other division methods in actual implementation.

[0164] The modules described as separate components may or may not be physically separated, and the components shown as modules may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0165] In addition, each functional module in each embodiment of the present invention may be integrated into one processing unit, each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of hardware plus software functional modules.

[0166] It is obvious to those skilled in the art that the present invention is not limited to the details of the above exemplary embodiments, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0167] Therefore, no matter from which point of view, the embodiments should be regarded as illustrative and non-restrictive, and the scope of the present invention is not limited only according to the above description, and it is intended that all changes within the meaning and scope of equivalent elements within the scope of protection are included in the present invention.

[0168] The embodiments of the present application can acquire and process relevant data based on artificial intelligence technology. Among them, artificial intelligence (AI) is the theory, method, technology and application system that uses digital computers or machines controlled by digital computers to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results.

[0169] In addition, it is obvious that the word "comprising" does not exclude other units or steps, and the singular does not exclude the plural. The multiple units or devices stated in the system embodiment can also be implemented by one unit or device through software or hardware. The words first, second, etc. are used to indicate names, and do not indicate any specific order.

[0170] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solution of the present invention can be modified or replaced by equivalents without departing from the spirit and scope of the technical solution of the present invention.

Claims

1. A fabric AI quality inspection method, characterized in that: The method comprises: Collecting image data of the target fabric, performing image enhancement processing on the image data to obtain image enhancement data, and extracting image quality inspection features corresponding to the image enhancement data; Dividing the image enhancement data into regions to obtain image block regions of the target fabric, and identifying multi-dimensional quality inspection elements corresponding to the image block regions one by one according to the image quality inspection features; Detecting a multidimensional defect element corresponding to the image block region according to the multidimensional quality inspection element, and determining a quality inspection factor corresponding to the image block region according to the multidimensional defect element; Marking the defective area of ​​the target fabric according to the quality inspection factor, and calculating the defect value corresponding to the defective area by using a preset multi-dimensional dynamic quality inspection algorithm, including: counting the number of original defective elements corresponding to the defective area; identifying the number of real-time defective elements corresponding to the defective area according to a preset dynamic time factor; and calculating the defect value corresponding to the defective area according to the number of original defective elements and the number of real-time defective elements by using the following preset multi-dimensional dynamic quality inspection algorithm: Among them, Q i is the defect value corresponding to the i-th defect area, s i is the severity factor of the i-th defect area, d0 is the number of original defect elements, max is the maximum value function, d it is the number of real-time defective elements in the i-th defective area at the t-th time, and T is the total time; The quality inspection value of the target fabric is determined by the defect value, and the quality inspection status of the target fabric is generated according to the quality inspection value.

2. The fabric AI quality inspection method according to claim 1, characterized in that: The performing image enhancement processing on the image data to obtain image enhancement data comprises: Performing denoising processing on the image data to obtain denoised image data; Performing contrast adjustment on the denoised image data; Performing sharpening processing on the denoised image data after contrast adjustment; The denoised image data after sharpening is subjected to color correction to obtain image enhancement data.

3. The fabric AI quality inspection method according to claim 1, characterized in that: The extracting the image quality inspection features corresponding to the image enhancement data includes: Generating a color histogram corresponding to the image enhancement data; Calculating the color distribution of the image enhancement data according to the color histogram; Extracting color quality inspection features corresponding to the image enhancement data according to the color distribution; Identifying crack features corresponding to the image enhancement data; Extracting damage quality inspection features corresponding to the image enhancement data according to the crack features; The color quality inspection feature and the damage quality inspection feature are merged into an image quality inspection feature.

4. The fabric AI quality inspection method according to claim 1, characterized in that: The step of dividing the image enhancement data into regions to obtain the image block regions of the target fabric includes: Counting the total number of pixels corresponding to the image enhancement data; Generate a division unit corresponding to the image enhancement data according to the total number of pixels; The image enhancement data is divided into regions according to the division unit to obtain image block regions of the target fabric.

5. The fabric AI quality inspection method according to claim 1, characterized in that: The step of identifying the multi-dimensional quality inspection elements corresponding to the image block regions one by one according to the image quality inspection features includes: Identify the color quality inspection sub-features and damage quality inspection sub-features corresponding to the image block areas one by one; Comparing the color quality inspection feature in the image quality inspection feature with the color quality inspection sub-feature to obtain a quality inspection color element and a quality inspection stain element; Comparing the damage quality inspection feature in the image quality inspection feature with the damage quality inspection sub-feature to obtain a quality inspection damage element and a quality inspection texture element; The quality inspection color element, the quality inspection stain element, the quality inspection damage element and the quality inspection texture element are merged into a multi-dimensional quality inspection element corresponding to the image block area.

6. The fabric AI quality inspection method according to claim 1, characterized in that: The detecting the multi-dimensional defect element corresponding to the image block area according to the multi-dimensional quality inspection element includes: Obtain target multi-dimensional quality inspection elements of target quality inspection fabric; Comparing the multidimensional quality inspection elements with the target multidimensional quality inspection elements one by one to obtain comparison factors; The multi-dimensional defect factor corresponding to the image block area is determined according to the contrast factor.

7. The fabric AI quality inspection method according to claim 1, characterized in that: The determining the quality inspection factor corresponding to the image block area according to the multi-dimensional defect element includes: Counting the number of defect elements corresponding to the multi-dimensional defect elements corresponding to the image block area; When the number of defective elements is greater than a preset element number threshold, determining the quality inspection factor corresponding to the image block area as a preset first quality inspection factor; When the number of defective elements is equal to a preset element number threshold, the quality inspection factor corresponding to the image block area is determined as a preset second quality inspection factor.

8. The fabric AI quality inspection method according to claim 1, characterized in that: The step of marking the defective area of ​​the target fabric according to the quality inspection factor comprises: When the quality inspection factor is the first quality inspection factor, marking the image block area corresponding to the quality inspection factor as a defective area; When the quality inspection factor is the second quality inspection factor, the image block area corresponding to the quality inspection factor is marked as a non-defective area.

9. The fabric AI quality inspection method according to claim 1, characterized in that: Determining the quality inspection value of the target fabric by using the defect value includes: Determining the total area quantity according to the defective area and the non-defective area of ​​the target fabric; Determining the quality inspection feedback factor of the target fabric through a preset subjective requirement; The quality inspection value of the target fabric is calculated by the defect value, the total area number and the quality inspection feedback factor: Wherein, Z is the quality inspection value, δ j is the quality inspection feedback factor of the target fabric corresponding to the jth subjective user, Q i is the defect value corresponding to the i-th defect area, n is the number of defect areas, H is the total number of areas, γ is the number of element quality inspections of the target fabric, and m is the number of subjective users.

10. A fabric AI quality inspection device, characterized in that: The device comprises: An image quality inspection feature extraction module is used to collect image data of the target fabric, perform image enhancement processing on the image data to obtain image enhancement data, and extract image quality inspection features corresponding to the image enhancement data; A multi-dimensional quality inspection element identification module is used to divide the image enhancement data into regions to obtain image block regions of the target fabric, and identify multi-dimensional quality inspection elements corresponding to the image block regions one by one according to the image quality inspection features; A quality inspection factor determination module, configured to detect a multidimensional defect element corresponding to the image block region according to the multidimensional quality inspection element, and determine a quality inspection factor corresponding to the image block region according to the multidimensional defect element; A defect value calculation module, used for marking the defect area of ​​the target fabric according to the quality inspection factor, and calculating the defect value corresponding to the defect area by using a preset multi-dimensional dynamic quality inspection algorithm; The quality inspection status generating module is used to determine the quality inspection value of the target fabric according to the defect value, and to generate the quality inspection status of the target fabric according to the quality inspection value.

Images