Visual identification method and system for dyeing defects of a blended fabric

Abstract

The present application relates to the technical field of image data processing, in particular to a kind of blended fabric dyeing defect visual identification method and system.The method comprises: under multiple different preset exposure times, multiple frames of exposure images are collected respectively, and multiple target images are obtained by affine transformation to multiple frames of exposure images;Based on multiple target images, the brightness change of each image position in multiple target images is analyzed, and the exposure response characteristics corresponding to each image position are obtained;The target brightness change rate corresponding to each image position and the corresponding saturation cutoff deviation are fused to obtain the degree of float color defect value corresponding to each image position;According to the degree of float color defect value corresponding to multiple image positions, dyeing process is graded adjustment.The present application can improve the visual identification effect of blended fabric dyeing defect.

Description

Technical Field

[0001] This invention relates to the field of image data processing technology, and specifically to a visual recognition method and system for dyeing defects in blended fabrics. Background Technology

[0002] In the textile processing industry, blended fabrics are widely used in apparel, home textiles, household goods, and industrial textiles due to their excellent overall performance and wide range of applications. Blended fabrics are woven from two or more different fiber raw materials. The molecular structure, physicochemical properties, and dyeing characteristics of different fibers vary significantly, resulting in a complicated dyeing process and stringent process control requirements. Fluctuations in parameters can easily lead to dyeing defects such as uneven dyeing and floating dye, making defect control difficult and impossible to completely avoid during production.

[0003] Currently, common methods for detecting floating color defects in blended fabrics include washing tests, color fastness sampling tests, and single visual color difference tests. Washing and color fastness tests qualitatively assess the floating color condition and dye binding stability through experimental methods, while traditional visual inspection relies on red-green-blue (RGB) color difference comparisons to roughly determine defects. Existing tests are mostly offline sampling inspections, which cannot identify localized hidden floating color in real time, resulting in a high risk of missed detections. Single color difference tests are easily affected by lighting, texture, and background color, leading to large errors and difficulty in distinguishing between normal color differences and floating color defects. At the same time, existing methods lack quantitative indicators and grading mechanisms, failing to provide an effective basis for precise control of the dyeing process.

[0004] In other words, the visual recognition effect of dyeing defects in blended fabrics using existing technologies is poor. Summary of the Invention

[0005] The purpose of this invention is to provide a visual recognition method and system for dyeing defects in blended fabrics, which solves the technical problem of poor visual recognition effect of dyeing defects in blended fabrics in the prior art.

[0006] In a first aspect, one embodiment of the present invention provides a visual identification method for dyeing defects in blended fabrics, the method comprising: Multiple frames of exposure images were acquired at different preset exposure times, and affine transformations were performed on the multiple frames of exposure images to obtain multiple target images. Based on multiple target images, the brightness change of each image location in the multiple target images is analyzed to obtain the exposure response features corresponding to each image location. The exposure response features include the target brightness change rate and the saturation cutoff deviation. The target brightness change rate is used to characterize the degree of difference in the change of exposure response between the floating color area and the normal area, and the saturation cutoff deviation is used to characterize the degree of nonlinear deviation caused by oversaturation in the floating color area. The target brightness change rate and the corresponding saturation cutoff deviation for each image location are fused to obtain the floating color defect degree value for each image location. The floating color defect degree value is used to characterize the severity of floating color defects in the fabric area. The target brightness change rate is positively correlated with the floating color defect degree value, and the saturation cutoff deviation is positively correlated with the floating color defect degree value. The dyeing process is graded and adjusted based on the degree of floating color defects corresponding to multiple image locations.

[0007] In one embodiment, the floating color defect value corresponding to each image position is calculated by weighting the target brightness change rate and the saturation truncation deviation corresponding to each image position.

[0008] In one embodiment, the step of analyzing the brightness change of each image location in multiple target images to obtain the exposure response features corresponding to each image location includes: From multiple target images, sequentially extract the brightness data corresponding to the same image location at different exposure times to obtain the brightness sequence corresponding to each image location; By performing fitting analysis on the brightness sequence corresponding to each image location, the target brightness change rate and the corresponding saturation cutoff deviation for each image location are obtained.

[0009] In one embodiment, the fitting analysis based on the brightness sequence corresponding to each image location to obtain the target brightness change rate and the corresponding saturation cutoff deviation for each image location includes: Curve fitting is performed on the brightness sequence corresponding to each image location to obtain the exposure response curve corresponding to each image location; A weighted global linear fit is performed on the exposure response curve corresponding to each image location to obtain the fitted straight line; Analyze the difference between the exposure response curve and the fitted straight line corresponding to each image position to obtain the saturation cutoff deviation corresponding to each image position; The slope of the fitted line corresponding to each image position is determined as the initial brightness change rate corresponding to each image position. Based on the saturation truncation deviation corresponding to each image position, the corresponding initial brightness change rate is corrected to obtain the target brightness change rate corresponding to each image position.

[0010] In one embodiment, the step of correcting the initial brightness change rate based on the saturation truncation deviation corresponding to each image position to obtain the target brightness change rate corresponding to each image position includes: A nonlinear correction function is constructed, and the saturation truncation deviation of each image position is substituted into the nonlinear correction function as an input variable to calculate the corresponding nonlinear correction coefficient. The initial brightness change rate corresponding to each image position is multiplied by the nonlinear correction coefficient to obtain the target brightness change rate corresponding to each image position. The nonlinear correction function is configured such that: when the saturation cutoff deviation is greater than a preset deviation threshold, the nonlinear correction coefficient is greater than 1 and increases nonlinearly with the increase of the saturation cutoff deviation; when the saturation cutoff deviation is less than or equal to the preset deviation threshold, the nonlinear correction coefficient is equal to 1.

[0011] In one embodiment, fusing the target brightness change rate and the corresponding saturation truncation deviation for each image location to obtain the floating color defect degree value for each image location includes: The target brightness change rate and saturation truncation deviation corresponding to each image position are normalized to obtain the brightness normalization value and saturation normalization value corresponding to each image position. The weighted sum of the normalized brightness and normalized saturation values ​​for each image location yields the degree of color floating defects for each image location.

[0012] In one embodiment, before performing a weighted summation of the normalized brightness value and normalized saturation value corresponding to each image location to obtain the floating color defect degree value corresponding to each image location, the method further includes: The average gray value of each image location in multiple target images is calculated by averaging the gray values ​​of each image location. Based on the interval of the average gray value of each image location, the initial calculation weight of the brightness normalization value corresponding to each image location is determined, wherein the initial calculation weight is positively correlated with the center value of the corresponding interval. The target calculation weight of the saturation normalization value corresponding to each image position is determined based on the initial calculation weight of the brightness normalization value corresponding to each image position. The sum of the initial calculation weight and the target calculation weight is 1. The saturation normalization value participates in the weighted summation process based on its corresponding target calculation weight.

[0013] In one embodiment, the step of adjusting the dyeing process according to the degree of floating color defects corresponding to multiple image locations includes: Establish a mapping relationship between the floating color defect level and the threshold range of the floating color defect degree value, wherein the floating color defect level includes at least two levels; Based on the mapping relationship, the floating color defect degree values ​​corresponding to multiple image locations are mapped to the corresponding floating color defect levels; The dyeing process is adjusted according to the level of floating color defects.

[0014] In one embodiment, performing an affine transformation on multiple exposed images to obtain multiple target images includes: A reference image is obtained, and the affine transformation matrix is ​​calculated based on the feature point matching relationship between the reference image and the other frames in the multi-frame exposure images. Using the affine transformation matrix, geometric correction and registration are performed on the corresponding multi-frame exposure images to obtain multiple target images.

[0015] Secondly, another embodiment of the present invention provides a visual recognition system for dyeing defects in blended fabrics, the system comprising: The preprocessing module is used to acquire multiple frames of exposure images at multiple different preset exposure times, and to perform affine transformation on the multiple frames of exposure images to obtain multiple target images; The analysis module is used to analyze the brightness change of each image position in multiple target images to obtain the exposure response features corresponding to each image position. The exposure response features include the target brightness change rate and the saturation cutoff deviation. The target brightness change rate is used to characterize the difference in exposure response between the floating color area and the normal area, and the saturation cutoff deviation is used to characterize the degree of nonlinear deviation caused by oversaturation in the floating color area. The fusion module is used to fuse the target brightness change rate and the corresponding saturation cutoff deviation for each image location to obtain the floating color defect degree value for each image location. The floating color defect degree value is used to characterize the severity of floating color defects in the fabric area. The target brightness change rate is positively correlated with the floating color defect degree value, and the saturation cutoff deviation is also positively correlated with the floating color defect degree value. The adjustment module is used to grade and adjust the dyeing process according to the degree of floating color defects corresponding to multiple image positions.

[0016] Thirdly, in another embodiment of the present invention, an electronic device is provided, including a processor, a memory, and a computer program stored in the memory and executable on the processor, wherein the computer program, when executed by the processor, implements the steps of the method described in the first aspect.

[0017] Fourthly, in another embodiment of the present invention, a computer-readable storage medium is provided, on which a computer program is stored, which, when executed by a processor, implements the steps of the method described in the first aspect.

[0018] The present invention has the following beneficial effects: This invention effectively overcomes positional deviations during image acquisition by acquiring multiple exposure images and aligning them using affine transformation, ensuring the accuracy of subsequent analysis. By analyzing the brightness changes of each image location across multiple frames, two key features—the target brightness change rate and saturation cutoff deviation—are extracted. This allows for precise quantification of the differences in exposure response between the floating color area and the normal area, as well as the nonlinear deviation caused by oversaturation. Fusing these two positively correlated features generates a floating color defect severity value, enabling a quantitative assessment of the severity of floating color defects in fabrics and providing reliable data for the graded adjustment of the dyeing process. This method improves the accuracy and objectivity of floating color defect detection, helps to promptly identify and correct dyeing problems, improves product quality and production efficiency, and ultimately enhances the visual recognition effect of dyeing defects in blended fabrics. Attached Figure Description

[0019] To more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 This is a flowchart illustrating a visual identification method for dyeing defects in blended fabrics provided in an embodiment of the present invention. Figure 2 This is a schematic diagram of the structure of a visual recognition system for dyeing defects in blended fabrics provided in an embodiment of the present invention; Figure 3 This is a schematic diagram of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0021] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of a visual identification method and system for dyeing defects in blended fabrics proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0022] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0023] The following description, in conjunction with the accompanying drawings, details a specific scheme for a visual identification method for dyeing defects in blended fabrics provided by the present invention.

[0024] This invention proposes a visual identification method for dyeing defects in blended fabrics. Please refer to [link / reference]. Figure 1 The diagram illustrates a schematic flowchart of a visual identification method for dyeing defects in blended fabrics according to an embodiment of the present invention. The method includes the following steps: Step S1: Acquire multiple frames of exposure images at multiple different preset exposure times, and perform affine transformation on the multiple frames of exposure images to obtain multiple target images.

[0025] The preset exposure time refers to multiple fixed exposure durations set in advance. For example, six fixed exposure times can be set: t1=1ms, t2=1.5ms, t3=2ms, t4=2.5ms, t5=3ms, and t6=4ms, to collect different brightness responses of the same fabric area.

[0026] Multi-frame exposure images refer to multiple frames of images continuously acquired for the same detection area of ​​a blended fabric according to multiple preset sets of different exposure times. These images contain fabric brightness response information under different exposure parameters, providing raw data for pixel-level exposure difference analysis and brightness change rate calculation.

[0027] Affine transformation is used to perform pixel-level alignment of images acquired at different exposure times. Specifically, image features can be extracted and feature matching can be completed through scale-invariant feature transformation (SIFT). Then, the random sample consensus (RANSAC) algorithm is used to eliminate mismatched pairs. The affine transformation matrix is ​​calculated and the transformation is performed to eliminate the positional offset caused by slight camera vibrations and fabric movements, so that the pixel coordinates of multi-exposure images can be matched one-to-one.

[0028] Multiple target images refer to multiple standard images that have undergone pixel-level alignment through affine transformation.

[0029] It should be noted that, in the embodiments of the present invention, multiple sets of different exposure times are preset, and multiple frames of original exposure images are acquired sequentially according to the preset exposure time for the same detection area of ​​the blended fabric. Due to camera vibration or slight fabric displacement during the acquisition process, pixel-level offsets may occur between multiple frames of images. Therefore, affine transformation is used to perform pixel-level registration and alignment of each frame of original exposure images to eliminate inter-frame positional deviations and ensure that the pixel coordinates of all images remain in one-to-one correspondence. The multiple standardized images after alignment processing are multiple target images, which are used to extract brightness data under multiple exposures pixel by pixel, construct brightness sequences, and conduct exposure response feature analysis.

[0030] Furthermore, the step of performing an affine transformation on multiple exposed images to obtain multiple target images includes: A reference image is obtained, and the affine transformation matrix is ​​calculated based on the feature point matching relationship between the reference image and the other frames in the multi-frame exposure images.

[0031] Among them, the reference image refers to the image acquired with the shortest exposure time in the multi-exposure image acquisition sequence as the reference image. It serves as a unified reference for feature matching, geometric correction and registration of all subsequent multi-frame exposure images. It is used to fix the image coordinate system and eliminate the positional offset caused by equipment vibration and timing deviation during the acquisition process.

[0032] For example, the feature point matching relationship refers to the stable and accurate pixel-level correspondence between the reference image and each exposure image established by extracting stable key points and feature descriptors of the reference image and each multi-frame exposure image through the SIFT feature detection algorithm, completing feature pairing with the K-nearest neighbor algorithm, and removing erroneous matching pairs through the RANSAC algorithm.

[0033] The affine transformation matrix is ​​a mathematical matrix calculated based on the matching relationship of feature points. It is used to describe the spatial position transformation relationship between images, such as translation, rotation, and perspective shift. It can accurately quantify the image offset and provide a mathematical basis for geometric correction and registration.

[0034] Using the affine transformation matrix, geometric correction and registration are performed on the corresponding multi-frame exposure images to obtain multiple target images.

[0035] Among them, geometric correction and registration refers to using an affine transformation matrix to perform pixel-level remapping processing on multiple frames of exposed images, eliminating positional offsets caused by production line vibrations and camera acquisition timing differences, so that each frame of exposed images is aligned with the reference image at the pixel level, ensuring that the same physical location corresponds to the same pixel coordinates in different exposed images.

[0036] The target image refers to a multi-frame exposure image that has been geometrically corrected and registered to the reference image, achieving pixel-level alignment and no positional deviation. It serves as the standard input image for subsequent exposure difference analysis, brightness change rate calculation, color floating defect identification, and grading.

[0037] Step S2: Based on multiple target images, analyze the brightness change of each image position in the multiple target images to obtain the exposure response features corresponding to each image position. The exposure response features include the target brightness change rate and the saturation cutoff deviation. The target brightness change rate is used to characterize the degree of difference in exposure response between the floating color area and the normal area. The saturation cutoff deviation is used to characterize the degree of nonlinear deviation caused by oversaturation in the floating color area.

[0038] Among them, the image position refers to a single pixel point in the target image identified by pixel coordinates, corresponding to a single detection point on the surface of the blended fabric, and is the basic unit for performing point-by-point brightness change analysis and exposure response feature calculation.

[0039] For example, the image position is the smallest unit for calculating the floating color defect feature. The corresponding feature point pairs need to be obtained through SIFT feature detection and matching. The RANSAC algorithm is used to remove mismatched points and calculate the affine transformation matrix. Based on this matrix, the pixel-level alignment of the multi-exposure images is completed. The alignment error is no more than 1 pixel, ensuring that the defect degree value accurately corresponds to the actual physical position of the fabric.

[0040] Brightness variation refers to the amount and rate of change of gray value corresponding to the same image position under different exposure times. It is used to reflect the influence of the dye distribution state on the fabric surface on the light reflection characteristics and is the basis for distinguishing floating color defects from normal dyeing areas.

[0041] For example, brightness variation refers to the increase or decrease of grayscale value at the same image location (pixel) under different exposure times; the brightness of normally stained areas increases linearly and steadily with exposure time, without obvious abrupt changes.

[0042] The floating color area refers to the defective area on the surface of the blended fabric where the dye has not fully penetrated and fixed, and has only accumulated in a free state. This area has high light reflection intensity, fast exposure saturation speed, and obvious abnormalities in brightness change and response curve.

[0043] Exposure response characteristics are composite feature parameters used to quantify the change of image position brightness with exposure time and to distinguish between over-color areas and normal areas. They consist of the target brightness change rate and saturation cutoff deviation, and can stably reflect the exposure saturation anomalies and nonlinear response characteristics caused by over-color defects.

[0044] For example, the floating color area has a strong scattering due to the aggregation of dye particles, and its reflectivity is significantly higher than that of the normal dyeing area. Under the same exposure conditions, it is easier to reach the saturation threshold of the photosensitive element more quickly, so that its exposure response characteristics are significantly different from those of the normal area. By using dynamic weight fusion processing based on the brightness of the background color, the degree value of floating color defect is obtained, so as to realize the graded judgment of floating color defect.

[0045] The target brightness change rate refers to the comprehensive brightness change rate obtained by weighted linear fitting and nonlinear correction of the exposure response curve of the same image position. It is used to characterize the theoretical brightness increase rate of the image position per unit exposure time under no saturation interference. It can distinguish the difference in the rate of exposure response between the floating color area and the normal area. The floating color area has a larger target brightness change rate because it has higher reflectivity and faster saturation, while the brightness change rate of the normal area tends to be stable.

[0046] It should be noted that, in the embodiments of the present invention, the nonlinear correction coefficient is calculated based on the saturation cutoff deviation. After correcting the initial brightness change rate, the target brightness change rate can be obtained. The target brightness change rate can reflect the exposure saturation mutation characteristics of the floating color area. After weighted fusion with the saturation cutoff deviation, the saturation nonlinearity and background interference can be suppressed, the true dyeing state of the fabric can be stably characterized, and the floating color defect degree value can be obtained, providing a basis for defect identification and classification.

[0047] Saturation cutoff deviation refers to the degree to which the exposure response curve of an image location deviates from the ideal linear response model (fitted straight line). It is used to quantitatively characterize the nonlinear deviation caused by premature saturation of the overprinted color area due to dye accumulation and excessive reflectivity. The greater the deviation, the more severe the overprinted color accumulation and the more significant the nonlinear characteristics of the response curve.

[0048] The normal area refers to the defect-free area on the surface of the blended fabric where the dye has fully penetrated and is uniformly fixed. The brightness of this area changes approximately linearly with exposure time, and the response curve shows no obvious non-linear deviation.

[0049] Furthermore, the step of analyzing the brightness change of each image location in multiple target images to obtain the exposure response features corresponding to each image location includes: Brightness data corresponding to the same image location at different exposure times are sequentially extracted from multiple target images to obtain the brightness sequence corresponding to each image location.

[0050] Different exposure times refer to the duration for which the photosensitive element receives incident light when the industrial camera acquires the target image. For example, multiple fixed exposure times can be set, and the characteristic that the accumulation of dye in the floating dye area leads to faster exposure saturation can be utilized. Gradient exposure can make the floating dye area and the normally dyed area present different brightness responses, providing conditions for defect identification.

[0051] Brightness data refers to the grayscale value of a single pixel in the target image after noise reduction, representing the intensity of reflected light at that point. The brightness data ranges from 0 to 255. The brightness of normally dyed areas changes stably and linearly with exposure, while areas with floating color defects exhibit high brightness due to the strong scattering effect of dye and quickly reach saturation cutoff.

[0052] A brightness sequence is a set of gray values ​​arranged in chronological order at the same image location under multiple exposures. A brightness sequence can fully reflect the brightness response law of a pixel as exposure changes. The brightness sequence in a normal area is approximately linearly distributed, while the brightness sequence in an area with color floating defects exhibits nonlinear saturation truncation characteristics.

[0053] For example, the brightness of the normally dyed area increases linearly and steadily with exposure; the brightness of the floating dye area rises rapidly and enters saturation prematurely due to the high reflectivity of the dye accumulation, and the sequence exhibits a significant nonlinear truncation characteristic.

[0054] By performing fitting analysis on the brightness sequence corresponding to each image location, the target brightness change rate and the corresponding saturation cutoff deviation for each image location are obtained.

[0055] Among them, fitting analysis refers to the process of using exposure time as the independent variable and gray value as the dependent variable to perform weighted linear regression fitting on the brightness sequence, and obtaining the exposure response curve and fitting parameters.

[0056] Furthermore, the fitting analysis based on the brightness sequence corresponding to each image location to obtain the target brightness change rate and the corresponding saturation cutoff deviation for each image location includes: Curve fitting is performed on the brightness sequence corresponding to each image location to obtain the exposure response curve for each image location.

[0057] Among them, the exposure response curve of the normal area is close to an ideal straight line; the exposure response curve of the floating color area is curved and truncated due to rapid saturation, and the degree of nonlinearity is positively correlated with the severity of floating color.

[0058] A weighted global linear fit is performed on the exposure response curve corresponding to each image location to obtain the fitted straight line.

[0059] Linear fitting refers to the process of using a linear model to perform overall optimal fitting on all exposure points of the brightness sequence; for example, by using weighted global linear fitting, the weight of saturation points with gray values ​​≥240 is reduced to 0.1 to avoid saturation points dominating the fitting results and to ensure that the fitted line can reflect the ideal linear response law without interference.

[0060] The fitted line refers to the optimal line output by the global linear fitting, representing the theoretical brightness response benchmark under conditions of no saturation and no floating color interference.

[0061] By analyzing the difference between the exposure response curve and the fitted straight line at each image location, the saturation cutoff deviation at each image location is obtained.

[0062] For example, the saturation cutoff deviation can be quantified by the distance or degree of discrete deviation between the exposure response curve and the linear fitting line; for example, the vertical distance between the exposure response curve and the linear fitting line can be calculated point by point, and the weighted square sum of the distances of all exposure points can be obtained to obtain the overall deviation; or the cumulative difference between the fitted line and the actual response points can be calculated based on discrete points. The larger the difference, the more significant the deviation of the actual response from linearity and the more severe the saturation cutoff.

[0063] The slope of the fitted line corresponding to each image location is determined as the initial brightness change rate corresponding to each image location.

[0064] The initial brightness change rate is used to characterize the theoretical brightness increase rate of an image position per unit exposure time under conditions of no saturation interference.

[0065] It should be noted that the initial brightness change rate is an uncorrected basic characteristic, which is easily affected by the depth of the fabric's base color and the texture of the blended fabric. It cannot accurately distinguish between normal dyeing differences and floating color defects, and needs to be corrected by saturation cutoff deviation.

[0066] It should be noted that, in the embodiments of the present invention, a weighted global linear fitting is performed on the exposure response curve corresponding to each image position. During the fitting process, saturated sampling points with gray values ​​greater than or equal to the saturation threshold are given lower weights or subjected to weight reduction processing, so that their influence on the fitting result is less than that of unsaturated sampling points. This eliminates the interference of the later saturation cutoff region on the fitting slope, and obtains a fitting straight line characterizing the theoretical brightness growth rate under no saturation interference. The slope of this straight line is then determined as the initial brightness change rate.

[0067] Based on the saturation truncation deviation corresponding to each image position, the corresponding initial brightness change rate is corrected to obtain the target brightness change rate corresponding to each image position.

[0068] For example, the target brightness change rate and saturation cutoff deviation are both processed using the Min-Max Normalization method, which maps the feature values ​​uniformly to the range of 0 to 1. For instance, the feature value range of a normal, defect-free fabric sample is used as the normalization benchmark, and the minimum and maximum values ​​of the target brightness change rate and the saturation cutoff deviation within the normal area are taken as the boundary values ​​for the normalization calculation, thereby eliminating the influence of differences in the dimensions and numerical ranges of different features on the fusion results.

[0069] The floating color defect level value refers to a value of 0 to 1 obtained by weighting and fusing the target brightness change rate after maximum-minimum normalization with the saturation cutoff deviation according to the dynamic weight corresponding to the brightness of the fabric background color. The larger the value, the more serious the floating color defect. It is used for defect area extraction, grade classification and production line process adjustment.

[0070] For example, the degree value of color floating defects It can be represented as: in, Indicates the initial weight calculation. Indicates the normalized value of brightness. Indicates the target weight calculation. This represents the saturation normalization value.

[0071] For example, the initial calculation weights It can be represented as: in, This represents the average grayscale value at each image location.

[0072] For example, the target is to calculate the weights. It can be represented as: in, This indicates the initial calculated weights.

[0073] The defect grading threshold refers to the critical value used to distinguish the degree of floating color defects, which is determined by the sample and is used to differentiate between no defects and light / moderate / severe floating color defects.

[0074] For example, no defects U < 0.2; slight defects 0.2 ≤ U < 0.5; moderate defects 0.5 ≤ U < 0.8; severe defects U ≥ 0.8, supporting real-time alarms and dynamic process adjustments on the production line.

[0075] Further, the step of correcting the initial brightness change rate based on the saturation truncation deviation corresponding to each image position to obtain the target brightness change rate corresponding to each image position includes: A nonlinear correction function is constructed, and the saturation truncation deviation of each image position is substituted into the nonlinear correction function as an input variable to calculate the corresponding nonlinear correction coefficient.

[0076] Among them, the nonlinear correction function refers to a monotonically increasing function that uses the saturation cutoff deviation as the only input variable to adaptively correct the initial brightness change rate. Its function is to suppress the interference of texture and lighting in the normal fabric area, significantly amplify the feature signal of the floating color defect area, and improve the robustness of floating color recognition.

[0077] The nonlinear correction coefficient refers to the scaling factor output by the nonlinear correction function, which is used to adaptively scale the initial brightness change rate; the coefficient for the normal area is 1, and the coefficient for the floating color area is greater than 1 and increases nonlinearly to enhance the defect features.

[0078] The target brightness change rate for each image location is obtained by multiplying the initial brightness change rate for each image location with the nonlinear correction coefficient.

[0079] Specifically, for each image location, the initial brightness change rate corresponding to that location is first obtained. This initial brightness change rate is the uncorrected basic change feature data. Then, a nonlinear correction coefficient adapted to the fabric imaging characteristics is introduced, and the initial brightness change rate of the same image location is multiplied with the nonlinear correction coefficient point by point to correct and compensate for errors such as background interference and imaging differences in the initial brightness change rate. After point-by-point correction calculation, the target brightness change rate corresponding to each image location is finally obtained, which effectively improves the accuracy and environmental adaptability of brightness change features and provides reliable basic data for subsequent multi-feature weighted fusion and accurate determination of floating color defects.

[0080] The nonlinear correction function is configured such that: when the saturation cutoff deviation is greater than a preset deviation threshold, the nonlinear correction coefficient is greater than 1 and increases nonlinearly with the increase of the saturation cutoff deviation; when the saturation cutoff deviation is less than or equal to the preset deviation threshold, the nonlinear correction coefficient is equal to 1.

[0081] Among them, the preset deviation threshold refers to the critical value statistically calibrated by a large number of normal blended fabric samples. It is used to distinguish the normal response area from the nonlinear distortion area that needs to be corrected, so as to avoid normal texture being misjudged as floating color defect.

[0082] The nonlinear increasing trend refers to the fact that the nonlinear correction coefficient increases rapidly as the deviation from the saturation cutoff increases. The more severe the color floating, the higher the magnification of the correction coefficient, which is consistent with the physical characteristics of accelerated exposure saturation caused by dye accumulation.

[0083] For example, nonlinear correction coefficient It can be represented as: in, This represents the nonlinear gain coefficient (empirically set to 4.0). Indicates the deviation from the saturation cutoff. This indicates the preset deviation threshold (based on experience, a value of 0.3 is used).

[0084] It should be noted that, in the embodiments of the present invention, when the saturation cutoff deviation... Greater than the preset deviation threshold When, correction factor Follow The increase of exhibits a quadratic nonlinear increase, highlighting the saturation distortion characteristics of the floating color region; when Less than or equal to When, correction factor Set to 1, and do not make any corrections.

[0085] Step S3: Fuse the target brightness change rate and the corresponding saturation cutoff deviation for each image position to obtain the floating color defect degree value for each image position. The floating color defect degree value is used to characterize the severity of floating color defects in the fabric area. The target brightness change rate is positively correlated with the floating color defect degree value, and the saturation cutoff deviation is positively correlated with the floating color defect degree value.

[0086] Fusion refers to the process of weighting the target brightness change rate and saturation cutoff deviation according to normalization and dynamic weight allocation to form a single comprehensive feature.

[0087] For example, the target brightness change rate and saturation cutoff deviation are first normalized to the range of 0 to 1. Then, the weights are adaptively assigned according to the brightness of the fabric background color. Finally, the fusion is completed by weighted summation, taking into account the advantages of both types of features and improving the anti-interference ability.

[0088] For example, the greater the rate of change of target brightness, the greater the degree of color floating defect; the greater the deviation of saturation cutoff, the greater the degree of color floating defect. Both features can individually reflect the severity of color floating, and their directions are consistent.

[0089] Furthermore, the floating color defect value corresponding to each image position is calculated by weighting the target brightness change rate and the saturation truncation deviation corresponding to each image position.

[0090] It should be noted that, in the embodiments of the present invention, the calculation weights of the target brightness change rate and the saturation cutoff deviation are dynamically adjusted according to the average gray value.

[0091] Among them, weighted calculation refers to the method of assigning dynamic weights to the target brightness change rate and saturation cutoff deviation based on the brightness of the fabric background color, and merging the two features into a value of floating color defect according to a preset formula.

[0092] The weight calculation refers to the proportional coefficient assigned to two features during weighted fusion. It is used to adjust the contribution of different features in defect determination, and the weight needs to be dynamically adjusted as the brightness of the fabric background color changes.

[0093] Furthermore, the process of fusing the target brightness change rate and the corresponding saturation truncation deviation for each image location to obtain the floating color defect degree value for each image location includes: The target brightness change rate and saturation truncation deviation corresponding to each image location are normalized to obtain the brightness normalization value and saturation normalization value corresponding to each image location.

[0094] Normalization refers to the process of mapping the rate of change of target brightness and the deviation of saturation cutoff from different units and numerical ranges to a unified range of 0 to 1 to standardize the values, so as to eliminate the differences in the numerical ranges of the two types of features and ensure fair comparability during weighted fusion.

[0095] The brightness normalization value refers to the standard value in the range of 0 to 1 obtained after the target brightness change rate is normalized by the maximum-minimum normalization process. The characteristic extreme value of the normalized sample of normal and defect-free fabric is used as the upper and lower limits of normalization to eliminate the influence of the dimension and numerical range. It is used to standardize and characterize the degree of brightness change. The larger the value, the more severe the brightness change of the pixel and the stronger the suspicion of color floating.

[0096] The saturation normalization value refers to the standard value in the range of 0 to 1 obtained after the saturation cutoff deviation is normalized by the maximum-minimum normalization process. The characteristic extreme value of the normalized sample is used as the upper and lower limits of the normalization to standardize and characterize the degree of nonlinear deviation of the exposure response curve. The larger the value, the more serious the deviation of the response curve from linearity and the more obvious the premature saturation of exposure. It is not affected by light and fabric texture, has strong stability, and is the core basis for judging floating color.

[0097] The weighted sum of the normalized brightness and normalized saturation values ​​for each image location yields the degree of color floating defects for each image location.

[0098] Among them, weighted summation refers to the operation of linearly weighting the brightness normalization value and the saturation normalization value according to the preset dynamic fusion weight.

[0099] It should be noted that, in the embodiments of the present invention, the weights are adaptively allocated based on the brightness of the fabric background color, with lighter colors emphasizing the brightness normalization value and darker colors emphasizing the saturation normalization value.

[0100] Furthermore, before performing a weighted summation of the normalized brightness and normalized saturation values ​​corresponding to each image location to obtain the floating color defect level value for each image location, the method further includes: The average gray value of each image location is calculated by averaging the gray values ​​across multiple target images.

[0101] Among them, grayscale value refers to the pixel brightness quantization value of an image position at a certain exposure time, with a value range of 0~255. The grayscale of the normally dyed area increases linearly with exposure, while the grayscale of the floating color area increases rapidly due to dye accumulation and is saturated and cut off.

[0102] The average gray value refers to the arithmetic mean of the gray values ​​of the same image location in all multiple exposed target images. It represents the overall reflective brightness level of that location and is used to determine the depth (light / medium / dark) of the fabric background color. It is the basis for dynamically allocating feature weights and eliminating the interference of background color differences on floating color recognition.

[0103] Based on the interval of the average gray value of each image location, the initial calculation weight of the brightness normalization value corresponding to each image location is determined, wherein the initial calculation weight is positively correlated with the center value of the corresponding interval.

[0104] Among them, the interval refers to the brightness level interval divided according to the average gray value, which is used to distinguish the fabric background color type and match the weight.

[0105] For example, the average grayscale value is divided into three levels: light ( ≥180), medium grayscale (80 < <180), dark color ( ≤80), different intervals correspond to different initial calculation weights.

[0106] The target calculation weight of the saturation normalization value corresponding to each image position is determined based on the initial calculation weight of the brightness normalization value corresponding to each image position. The sum of the initial calculation weight and the target calculation weight is 1. The saturation normalization value participates in the weighted summation process based on its corresponding target calculation weight.

[0107] The initial calculation weight refers to the weighting coefficient assigned to the brightness normalization value based on the interval where the average gray value lies. The initial calculation weight is positively correlated with the center value of the interval; the lighter the background color, the greater the weight (based on experience, a value of 0.9 is taken), and the darker the background color, the smaller the weight (based on experience, a value of 0.1 is taken), thus achieving adaptive feature emphasis.

[0108] The target calculation weight refers to the weighting coefficient assigned to the saturated normalized value, which is derived from the initial calculation weight. It satisfies the condition that the sum of the initial calculation weight and the target calculation weight is 1. The darker the background color, the larger the target calculation weight, complementing the initial calculation weight and jointly balancing the contributions of the two features.

[0109] Weighted summation refers to the process of linearly weighting the normalized brightness value and the normalized saturation value according to their respective weights to obtain the value of the degree of floating color defect.

[0110] Step S4: Adjust the dyeing process according to the degree of floating color defects corresponding to multiple image positions.

[0111] The dyeing process refers to the entire process of blended fabrics from dyeing, soaping, color fixing to drying, including key parameters such as dye concentration, temperature, time, tension, and soaping intensity.

[0112] For example, during the process of grading and adjusting the dyeing process, the focus is on controlling process parameters directly related to floating color, such as soaping time, fixing temperature, fixing agent dosage, and water washing flow rate, to suppress the generation of floating color from the source.

[0113] Graded adjustment refers to the differentiated and step-by-step correction and control of process parameters on the dyeing production line based on the level of floating color defects.

[0114] For example, minor floating color defects are finely adjusted; moderate defects are optimized in a targeted manner; and severe defects are immediately alarmed and significantly adjusted (such as soaping time +10% and color fixing temperature +5℃), thus achieving closed-loop control of defects and processes.

[0115] Furthermore, the step of adjusting the dyeing process in stages based on the degree of floating color defects corresponding to multiple image locations includes: Establish a mapping relationship between the floating color defect level and the threshold range of the floating color defect degree value, wherein the floating color defect level includes at least two levels.

[0116] Among them, the floating color defect level refers to the detection and judgment level divided according to the severity of the floating color defect, which is used to distinguish the severity of the defect and guide the process adjustment.

[0117] For example, the floating color defect levels include four levels: no defect, slight floating color defect, moderate floating color defect, and severe floating color defect, which can cover the entire range from qualified to seriously unqualified, meeting the needs of precise control of the production line.

[0118] For example, four levels of floating color defects are defined: no defects (floating color defect severity value < 0.2), slight floating color defects (0.2 ≤ floating color defect severity value < 0.5), moderate floating color defects (0.5 ≤ floating color defect severity value < 0.8), and severe floating color defects (floating color defect severity value ≥ 0.8). The grading thresholds are determined by comprehensive statistical calibration of 100 sets of normal and defective samples, effectively ensuring the stability of the grading standard and the reliability of the judgment results.

[0119] The threshold range refers to the numerical range defined by the samples and used to classify different levels of floating color defects.

[0120] The mapping relationship refers to the judgment rule that there is a one-to-one correspondence between the floating color defect level and the threshold range of the floating color defect value. That is, inputting the defect degree value of any pixel can directly output the corresponding level, realizing the automatic conversion from quantitative value to qualitative level and supporting automated grading.

[0121] Based on the mapping relationship, the floating color defect degree values ​​corresponding to multiple image locations are mapped to the corresponding floating color defect levels.

[0122] It should be noted that, in the embodiments of the present invention, based on the pre-established numerical and grade mapping relationship, threshold comparison and matching judgment are completed one by one according to the floating color defect degree value corresponding to each image position; the floating color defect degree value obtained by quantization of different pixel positions is mapped to the corresponding floating color defect level, realizing the standardized conversion from quantized value to defect level, thereby completing the unified classification of floating color defects in the entire fabric area, providing a clear grade basis for subsequent defect area labeling, quality assessment and process optimization.

[0123] The dyeing process is adjusted according to the level of floating color defects.

[0124] For example, based on the floating color defect level determined in each region, targeted graded control and differentiated adjustments are carried out in the dyeing process. For floating color defects of different levels—no defects, mild, moderate, and severe—corresponding process optimization strategies are set, and key production conditions such as dyeing raw material ratio, reaction temperature, dyeing time, and fixing parameters are adjusted as needed. Through a refined control approach of graded measures, the uniformity of fabric dyeing and the state of dye binding are improved, the generation of floating color defects is suppressed from the source of production, and the closed-loop optimization of the dyeing process and the stable improvement of product dyeing quality are achieved.

[0125] In summary, this invention effectively overcomes positional bias during image acquisition by acquiring multiple exposure images and aligning them using affine transformation, ensuring the accuracy of subsequent analysis. By analyzing the brightness changes of each image location across multiple frames, two key features—the target brightness change rate and saturation cutoff deviation—are extracted. This allows for precise quantification of the differences in exposure response between the floating color area and the normal area, as well as the nonlinear deviation caused by oversaturation. Fusing these two positively correlated features generates a floating color defect severity value, enabling a quantitative assessment of the severity of floating color defects in fabrics and providing reliable data for the graded adjustment of the dyeing process. This method improves the accuracy and objectivity of floating color defect detection, helps to promptly identify and correct dyeing problems, improves product quality and production efficiency, and ultimately enhances the visual recognition effect of dyeing defects in blended fabrics.

[0126] This invention proposes a visual recognition system for dyeing defects in blended fabrics. Please refer to [link / reference]. Figure 2 The diagram illustrates a structural schematic of a visual recognition system 200 for dyeing defects in blended fabrics according to an embodiment of the present invention. The system includes: The preprocessing module 201 is used to acquire multiple frames of exposure images at multiple different preset exposure times, and to perform affine transformation on the multiple frames of exposure images to obtain multiple target images; Analysis module 202 is used to analyze the brightness change of each image position in multiple target images based on multiple target images, and obtain the exposure response features corresponding to each image position. The exposure response features include target brightness change rate and saturation cutoff deviation. The target brightness change rate is used to characterize the degree of difference in exposure response between the floating color area and the normal area, and the saturation cutoff deviation is used to characterize the degree of nonlinear deviation caused by oversaturation in the floating color area. The fusion module 203 is used to fuse the target brightness change rate and the corresponding saturation cutoff deviation for each image position to obtain the floating color defect degree value for each image position. The floating color defect degree value is used to characterize the severity of floating color defects in the fabric area. The target brightness change rate is positively correlated with the floating color defect degree value, and the saturation cutoff deviation is positively correlated with the floating color defect degree value. The adjustment module 204 is used to grade and adjust the dyeing process according to the floating color defect degree value corresponding to multiple image positions.

[0127] It should be noted that the system provided in the above embodiments is only an example of the division of the above functional modules. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the computer device can be divided into different functional modules to complete all or part of the functions described above. In addition, the visual recognition system for dyeing defects in blended fabrics and the visual recognition method for dyeing defects in blended fabrics provided in the above embodiments belong to the same concept, and their specific implementation process can be found in the method embodiments, which will not be repeated here.

[0128] This invention also provides an electronic device. Please refer to [link to relevant documentation]. Figure 3 The electronic device may include a processor 301, a memory 302, and a program 3021 stored in the memory 302 and capable of running on the processor 301.

[0129] When program 3021 is executed by processor 301, it can achieve the following: Figure 1 Any steps in the corresponding method embodiments and the achievement of the same beneficial effects will not be repeated here.

[0130] Those skilled in the art will understand that all or part of the steps of the methods described in the above embodiments can be implemented by hardware related to program instructions, and the program can be stored in a readable medium.

[0131] This invention also provides a readable storage medium storing a computer program, which, when executed by a processor, can perform the above-described functions. Figure 1 Any step in the corresponding method embodiment can achieve the same technical effect, and will not be repeated here to avoid repetition.

[0132] The computer-readable storage medium of this invention can be any combination of one or more computer-readable media. The computer-readable medium can be a computer-readable signal medium or a computer-readable storage medium. For example, a computer-readable storage medium can be an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of computer-readable storage media (a non-exhaustive list) include: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this document, a computer-readable storage medium can be any tangible medium that contains or stores a program that can be used by or in conjunction with an instruction execution system, apparatus, or device.

[0133] Computer-readable signal media may include data signals propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. Computer-readable signal media may also be any computer-readable medium other than computer-readable storage media, capable of sending, propagating, or transmitting programs for use by or in connection with an instruction execution system, apparatus, or device.

[0134] The program code contained on the storage medium can be transmitted using any suitable medium, including but not limited to wireless, wire, optical fiber, RF, etc., or any suitable combination thereof.

[0135] Computer program code for performing the operations of this invention can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, and C++, as well as conventional procedural programming languages ​​such as "C" or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or terminal. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0136] This invention also provides a computer program product that, when run on a computer, causes the computer to perform the aforementioned steps to achieve a visual identification method for dyeing defects in blended fabrics provided in the above embodiments.

[0137] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0138] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

Claims

1. A visual identification method for dyeing defects in blended fabrics, characterized in that, The method includes: Multiple frames of exposure images are acquired at various preset exposure times, and affine transformations are performed on the multiple frames of exposure images to obtain multiple target images. Based on the multiple target images, the brightness change of each image position in the multiple target images is analyzed to obtain the exposure response features corresponding to each image position. The exposure response features include the target brightness change rate and the saturation cutoff deviation. The target brightness change rate is used to characterize the degree of difference in exposure response between the floating color area and the normal area. The saturation cutoff deviation is used to characterize the degree of nonlinear deviation caused by oversaturation in the floating color area. The target brightness change rate and the corresponding saturation cutoff deviation for each image location are fused to obtain the floating color defect degree value for each image location. The floating color defect degree value is used to characterize the severity of floating color defects in the fabric area. The target brightness change rate is positively correlated with the floating color defect degree value, and the saturation cutoff deviation is positively correlated with the floating color defect degree value. The dyeing process is graded and adjusted based on the degree of floating color defects corresponding to multiple image locations.

2. The visual identification method for dyeing defects in blended fabrics according to claim 1, characterized in that, The floating color defect value corresponding to each image position is calculated by weighting the target brightness change rate and the saturation truncation deviation corresponding to each image position.

3. The visual identification method for dyeing defects in blended fabrics according to claim 1, characterized in that, The step of analyzing the brightness change of each image location in the multiple target images to obtain the exposure response features corresponding to each image location includes: From the multiple target images, sequentially extract the brightness data corresponding to the same image location at different exposure times to obtain the brightness sequence corresponding to each image location; By performing fitting analysis on the brightness sequence corresponding to each image location, the target brightness change rate and the corresponding saturation cutoff deviation for each image location are obtained.

4. The visual identification method for dyeing defects in blended fabrics according to claim 3, characterized in that, The fitting analysis based on the brightness sequence corresponding to each image location yields the target brightness change rate and the corresponding saturation cutoff deviation for each image location, including: Curve fitting is performed on the brightness sequence corresponding to each image location to obtain the exposure response curve corresponding to each image location; A weighted global linear fit is performed on the exposure response curve corresponding to each image location to obtain the fitted straight line; Analyze the difference between the exposure response curve corresponding to each image position and the fitted straight line to obtain the saturation cutoff deviation corresponding to each image position; The slope of the fitted line corresponding to each image position is determined as the initial brightness change rate corresponding to each image position. Based on the saturation truncation deviation corresponding to each image position, the corresponding initial brightness change rate is corrected to obtain the target brightness change rate corresponding to each image position.

5. The visual identification method for dyeing defects in blended fabrics according to claim 4, characterized in that, The step of correcting the initial brightness change rate based on the saturation truncation deviation at each image location to obtain the target brightness change rate at each image location includes: A nonlinear correction function is constructed, and the saturation truncation deviation of each image position is substituted into the nonlinear correction function as an input variable to calculate the corresponding nonlinear correction coefficient. The initial brightness change rate corresponding to each image position is multiplied by the nonlinear correction coefficient to obtain the target brightness change rate corresponding to each image position. The nonlinear correction function is configured such that: when the saturation cutoff deviation is greater than a preset deviation threshold, the value of the nonlinear correction coefficient is greater than 1 and increases nonlinearly with the increase of the saturation cutoff deviation; when the saturation cutoff deviation is less than or equal to the preset deviation threshold, the value of the nonlinear correction coefficient is equal to 1.

6. The visual identification method for dyeing defects in blended fabrics according to claim 1, characterized in that, The process of fusing the target brightness change rate and the corresponding saturation truncation deviation for each image location to obtain the floating color defect degree value for each image location includes: The target brightness change rate and the saturation truncation deviation corresponding to each image position are normalized to obtain the brightness normalization value and saturation normalization value corresponding to each image position. The weighted sum of the normalized brightness and normalized saturation values ​​for each image location yields the degree of color floating defects for each image location.

7. The visual identification method for dyeing defects in blended fabrics according to claim 6, characterized in that, Before performing a weighted summation of the normalized brightness and normalized saturation values ​​corresponding to each image location to obtain the floating color defect level value for each image location, the method further includes: The average gray value of each image location in multiple target images is calculated by averaging the gray values ​​of each image location. Based on the interval of the average gray value of each image location, the initial calculation weight of the brightness normalization value corresponding to each image location is determined, wherein the initial calculation weight is positively correlated with the center value of the corresponding interval; The target calculation weight of the saturation normalization value corresponding to each image position is determined based on the initial calculation weight of the brightness normalization value corresponding to each image position. The sum of the initial calculation weight and the target calculation weight is 1. The saturation normalization value participates in the weighted summation process based on its corresponding target calculation weight.

8. The visual identification method for dyeing defects in blended fabrics according to claim 1, characterized in that, The step of classifying and adjusting the dyeing process based on the degree of floating color defects corresponding to multiple image locations includes: Establish a mapping relationship between the floating color defect level and the threshold range of the floating color defect degree value, wherein the floating color defect level includes at least two levels; Based on the mapping relationship, the floating color defect degree values ​​corresponding to the multiple image positions are mapped to the corresponding floating color defect levels; The dyeing process is adjusted according to the level of floating color defect.

9. The visual identification method for dyeing defects in blended fabrics according to claim 1, characterized in that, The step of performing an affine transformation on the multi-frame exposure images to obtain multiple target images includes: A reference image is acquired, and an affine transformation matrix is ​​calculated based on the feature point matching relationship between the reference image and the remaining frames in the multi-frame exposure images. Using the affine transformation matrix, geometric correction and registration are performed on the corresponding multi-frame exposure images to obtain the multiple target images.

10. A visual recognition system for dyeing defects in blended fabrics, characterized in that, The system includes: The preprocessing module is used to acquire multiple frames of exposure images at multiple different preset exposure times, and to perform affine transformation on the multiple frames of exposure images to obtain multiple target images; The analysis module is used to analyze the brightness change of each image position in the multiple target images based on the multiple target images, and obtain the exposure response features corresponding to each image position. The exposure response features include the target brightness change rate and the saturation cutoff deviation. The target brightness change rate is used to characterize the degree of difference in the change of exposure response between the floating color area and the normal area. The saturation cutoff deviation is used to characterize the degree of nonlinear deviation caused by oversaturation in the floating color area. The fusion module is used to fuse the target brightness change rate and the corresponding saturation cutoff deviation for each image position to obtain the floating color defect degree value for each image position. The floating color defect degree value is used to characterize the severity of floating color defects in the fabric area. The target brightness change rate is positively correlated with the floating color defect degree value, and the saturation cutoff deviation is positively correlated with the floating color defect degree value. The adjustment module is used to grade and adjust the dyeing process according to the degree of floating color defects corresponding to multiple image positions.

Images