A method for identifying printing and dyeing processes in textile printing and dyeing

Abstract

This invention discloses a dyeing and printing identification method in textile dyeing and printing processes, belonging to the field of textile engineering technology. After the dyeing and printing process is completed, the method acquires textile images using an industrial area array camera placed at the end of the production line, extracts image fragment regions, and constructs a set of local dyeing features and a set of structural stability representations. Within the local analysis window, the image fragment regions are subdivided, and color distribution structural features and color transition trend features are extracted. These are then spatially aligned and correlated with texture direction features and structural morphology features to form a dyeing-structure feature set. Based on spatial correspondence, the dyeing response state and structural bearing characteristics are coupled and identified, classifying the detection area state into a structure-dyeing consistent state, a dyeing anomaly-dominated state, or a structural bearing anomaly-dominated state. Based on the identification type, the source of the anomaly is attributed and an anomaly cause label is attached, classifying and identifying the sources of dyeing and printing anomalies and labeling their regions.

Description

Technical Field

[0001] This invention relates to the field of textile engineering technology, specifically to a method for identifying dyeing processes in textile printing and dyeing. Background Technology

[0002] As the fundamental carrier of clothing, home textiles, and functional materials, textiles' appearance quality and structural state directly affect product grade and market application value. In the textile production process, dyeing and printing are crucial steps determining the fabric's color expression, pattern clarity, and overall visual consistency. The dyeing and printing process involves multiple steps, including sizing, dye penetration, color fixing, washing, and setting, which not only affect color distribution but also have a coupling effect on the fabric's structure, tension balance, and surface morphology. With the widespread adoption of automated production lines, how to simultaneously identify and determine the dyeing effect and structural state of the fabric after dyeing and printing has gradually become a core issue in quality control, thus creating a demand for dyeing and printing identification processes.

[0003] Existing dyeing and printing identification methods mostly focus on single-dimensional color detection, typically using color difference comparison or local grayscale statistics as the basis for judgment, rarely combining dyeing performance with fabric structural state for joint analysis. When uneven local color patches, blurred boundaries, or pattern shifts occur, existing methods often only indicate the presence of anomalies, failing to distinguish whether the anomalies are caused by uneven dye distribution or by structural factors such as fabric tension changes or texture deformation. Furthermore, in complex structural locations such as local splicing areas and edge finishing areas, analysis based solely on color or grayscale features is easily affected by changes in texture direction and lighting, resulting in insufficient identification stability and difficulty in clearly classifying the causes of anomalies. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides a method for identifying printing and dyeing processes in textile printing and dyeing, thus solving the problems mentioned in the background section.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a method for identifying printing and dyeing processes in textile printing and dyeing, comprising the following steps:

[0006] S1. After the dyeing and printing are completed, the textile images are acquired by an industrial area array camera. The target area of ​​the textile image is located according to the target area positioning rules, and a set of local dyeing performance and a set of structural stability performance are constructed.

[0007] S2. Based on the local staining feature set, the image segment region is divided into several sub-regions within the local analysis window, and the color distribution structure features and color transition trend features are extracted and correlated to form staining response state features.

[0008] S3. Within the local analysis window, based on the set of structural stability performance, the image segment region is scanned pixel by pixel according to the row and column order of each sub-region to extract texture direction features and structural morphology features. After spatial alignment in the same local analysis window coordinate system, the structural bearing features are evaluated.

[0009] S4. Construct a set of staining-structure features and perform staining-structure coupling identification. Divide the state of the detection area based on the identification results.

[0010] S5. Based on the staining-structure coupling identification results, perform attribution analysis on the source of the anomaly and attach corresponding anomaly cause labels.

[0011] Preferably, S1 includes S11;

[0012] S11. Under the same working conditions after the dyeing and printing process is completed, the textile is imaged by an industrial area array camera set at the end of the dyeing and printing production line. The target area of ​​the textile image is located according to the target area positioning rules. After the positioning is completed, a local analysis window is constructed with the spatial coordinate range of the target area as the boundary, and the corresponding image segment area of ​​the target area is extracted. The position identifier and boundary information of the current local analysis window in the image are recorded, and the image segment area is output.

[0013] The target area includes the cuff edge area, the neckline splicing area, and the pattern edge area.

[0014] Preferably, S1 further includes S12 and S13;

[0015] S12. Taking the image segment region as input, the original RGB data of the image is converted into CIELab color space data within the local analysis window. The color distribution density matrix is ​​calculated by performing partition histogram statistics and two-dimensional color co-occurrence matrix on the L, a, and b three-channel data. At the same time, the gradient of the color components of adjacent pixels is calculated by using the first-order difference operator to obtain the color transition gradient magnitude distribution. Then, the image color pixels are clustered based on the density clustering algorithm to extract the number of color clusters and the cluster area ratio. By performing variance and mean calculation on the pixel color values ​​within the local window, the color uniformity parameter is obtained, and a local coloring feature set is constructed.

[0016] S13. A grayscale image is generated by weighted summation of the RGB three-channel pixel values ​​of the image segment region. Gradient calculation is performed on the grayscale image to obtain the gradient direction and gradient magnitude of each pixel. The gradient direction data is statistically processed to count the frequency of occurrence and concentration of the main direction. Adjacent pixel pairs are set according to the preset pixel spacing and direction. The grayscale image is traversed pixel by pixel to count the occurrence of each grayscale level combination under the corresponding spacing and direction conditions. A grayscale level joint distribution table is constructed, and the joint distribution table is normalized to form a grayscale co-occurrence matrix. Based on the grayscale co-occurrence matrix, the concentration of grayscale difference distribution and the proportion of grayscale occurrence at the same level are calculated to obtain the texture grayscale change amplitude and texture arrangement consistency in the local area, and a structurally stable representation set is constructed.

[0017] Preferably, S2 includes S21, S22 and S23;

[0018] S21. Based on the local coloring feature set, the image segment region is divided into several sub-regions according to the image coordinates within the local analysis window. The mean chromaticity, chromaticity dispersion, and mean gradient amplitude are statistically calculated in each sub-region. The sub-regions are then arranged according to their spatial order to form color distribution structure features.

[0019] S22. Statistically analyze the gradient direction of each pixel, calculate the number of pixels in each direction interval, take the direction with the highest frequency as the main change direction, and traverse pixel by pixel along the main change direction, record the chromaticity difference change between consecutive pixels, when the chromaticity difference does not exceed the preset chromaticity threshold, record it as a gradient segment, when the chromaticity difference exceeds the preset chromaticity threshold, record it as a sudden change point, and merge consecutive sudden change points to determine the boundary position and generate color transition trend features.

[0020] S23. According to the sub-region number, associate the color distribution structure features of each sub-region with its corresponding color transition trend features to form a coloring response state feature that includes the region location, color distribution value, main direction angle, gradient length and boundary coordinates.

[0021] Preferably, S3 includes S31 and S32;

[0022] S31. Within the local analysis window, based on the set of structurally stable performance, the image segment region is scanned pixel by pixel according to the row and column order of each sub-region. The gradient direction difference and gradient magnitude difference between adjacent pixels are counted. When the gradient direction difference between adjacent regions is within a preset range, it is determined to be a texture continuous region, and its spatial range is recorded. At the same time, the number of texture lines per unit area and their distribution difference are counted in each sub-region. The texture density is calculated based on the local gradient field aggregation degree to form the texture direction characteristics of the image segment region.

[0023] S32. On the grayscale image, the grayscale values ​​are continuously sampled along the main change direction to obtain a one-dimensional grayscale sequence. The grayscale sequence is subjected to differential processing to calculate the grayscale change between adjacent sampling points. The amplitude range of grayscale change and the interval between adjacent peaks are statistically calculated according to the statistical method to form surface undulation data and structural morphological features.

[0024] Preferably, S3 further includes S33;

[0025] S33. Under the same local analysis window coordinate system, spatially align the boundary of the continuous texture region in the texture direction feature with the spacing offset position and boundary change position in the structural morphology feature, count the area ratio of the continuous texture region in the local analysis window, calculate the maximum value of the spacing offset value and the boundary position offset value and the length of the continuous over-limit segment in the continuous texture region, and then evaluate the structural bearing characteristics.

[0026] By dividing the statistical data of historical qualified samples and samples with structural anomalies under the same process conditions into intervals, the normal distribution range and abnormal distribution range of the corresponding parameters are calculated respectively, and the area threshold, spacing offset threshold and position offset threshold are set respectively at the boundary interval of the two types of sample distributions.

[0027] When the area ratio is not less than the preset area threshold, and the spacing offset value and the boundary position offset value do not exceed the corresponding spacing offset threshold and position offset threshold, output a structural stability status indicator.

[0028] When the area ratio is less than the preset area threshold, or the spacing offset value exceeds the corresponding spacing offset threshold, an abnormal structural status indicator is output.

[0029] When the area ratio is less than the preset area threshold and the boundary position offset value exceeds the position offset threshold, output a structural failure status indicator.

[0030] Preferably, S4 includes S41 and S42;

[0031] S41. Based on the principle of spatial consistency, the dyeing response state characteristics and structural bearing characteristics are mapped region by region, and the dyeing response state characteristics and structural bearing characteristics after region-by-region mapping are unified to the same identification scale through feature comparison analysis technology to form a dyeing-structure feature set.

[0032] S42. Based on the coloring-structure feature set, perform coloring-structure coupling identification, and the specific identification is as follows;

[0033] When the area of ​​abnormal dyeing does not overlap with the area of ​​structural bearing characteristics in space, it is determined that the abnormal dyeing is caused by dyeing process factors, indicating that the abnormal dyeing is the dominant state.

[0034] When the coloring anomaly area and the structural bearing characteristic area overlap in space, it is determined that the coloring anomaly is caused by the structural bearing condition limitation, indicating that the structural bearing anomaly is the dominant state.

[0035] When both the dyeing response state and the structural load-bearing state are stable within the overall space, the dyeing process is considered to be in a state of structural and technological matching, indicating a state of structural-dyeing consistency.

[0036] Preferably, S4 further includes S43;

[0037] S43. Analyze the staining-structure coupling identification results, convert the identification results into structured state information, and divide the current detection area state as follows;

[0038] When the dyeing abnormality is the dominant state, it means that the structural load-bearing state is within the normal range, but the dyeing performance is abnormally deviated. The abnormality comes from the dyeing process. At this time, the dyeing process adjustment suggestion is output and the relevant personnel are notified to make process corrections and improvements to the current area.

[0039] When structural load-bearing abnormalities dominate, it indicates that the dyeing is abnormal and the structural load-bearing capacity is insufficient. Continuing the dyeing process poses a quality risk. At this time, we will provide suggestions for adjusting the fabric tension and process the fabric condition before proceeding to the dyeing process.

[0040] When the structure-staining consistency is achieved, it indicates that there is consistency between the staining performance and the structural bearing condition, and no abnormal features are found. At this time, the current image segment region is judged to be in a qualified state and marked as a normal production sample.

[0041] Preferably, S5 includes S51;

[0042] S51. Based on the identification and transformation results, the current image segment region is identified by its region type, and the current image segment region is bound to the corresponding image segment region through the region positioning information to form a region category identifier.

[0043] The staining-structure coupling identification results are categorized based on the region category identifier. The identification results include the structure-staining consistency state, the staining abnormality-dominant state, and the structural bearing abnormality-dominant state, and the identification types are uniformly identified.

[0044] Preferably, S5 further includes S52;

[0045] S52. Based on the identification type, perform anomaly source attribution analysis on the current identified object and attach corresponding anomaly cause labels. The anomaly cause labels include dyeing process-related anomaly labels, structure-related anomaly labels, and non-process anomaly labels.

[0046] When the identification results indicate that the dyeing response is abnormal but the structural load-bearing state is stable, the current abnormality will be marked as a dyeing process-related abnormality.

[0047] When the identification results indicate that the coloring anomaly is related to the structural bearing state, the current anomaly is marked as a structurally related anomaly.

[0048] When the identification results show that the anomaly is not persistent and does not conform to the process characteristics, the current anomaly is marked as a non-process anomaly.

[0049] This invention provides a method for identifying dyeing processes in textile printing and dyeing. It has the following beneficial effects:

[0050] (1) This method processes image fragment regions under the same local analysis window coordinate system, extracting color distribution structure features, color transition trend features, texture direction features, and structural morphology features respectively, and further forming dyeing response state features and structural bearing features, realizing the partitioning modeling of dyed textiles at the microscopic spatial scale. Compared with detection methods based solely on overall color difference or single gray-level statistics, this scheme completes pixel-by-pixel scanning, gray-level gradient direction statistics, gray-level co-occurrence matrix calculation, and color clustering at the sub-region level, enabling the dyeing behavior and structural bearing state of local regions to have corresponding spatial expression forms, and completing the task of synchronously analyzing the dyeing appearance phenomenon and the fabric structural state.

[0051] (2) This method uses the principle of spatial consistency to correspond the dyeing response state features and structural bearing features region by region, and unifies them to the same identification scale to form a dyeing-structure feature set. Then, based on this set, dyeing-structure coupling identification is performed, realizing the differentiation and determination of the source of anomalies. This step no longer simply regards dyeing anomalies as color distribution anomalies, but combines the structural bearing state to analyze the spatial overlap relationship. Whether the dyeing anomaly area overlaps with the structural bearing feature area becomes the identification criterion, so that the detection results have structural correlation logic. By distinguishing the structure-dyeing consistent state, the dyeing anomaly-dominant state, and the structural bearing anomaly-dominant state, the transformation from judging the existence of anomalies to judging the dominant factors of anomalies is completed. Compared with the existing single-channel visual detection technology, this scheme completes the fusion of dual feature systems within the same local analysis window, so that the identification results have a spatial correlation basis, rather than isolated statistical results.

[0052] (3) This method identifies the region category based on the identification and transformation results, and binds the current image segment region with the corresponding image segment region through the region positioning information. At the same time, it performs anomaly source attribution analysis based on the identification type and adds anomaly cause labels, realizing the structured output of the identification results. The dyeing-structure coupling identification results are mapped with the actual production area, so that each abnormal state has a clear region category identifier and anomaly cause label, including dyeing process-related anomaly labels, structure-related anomaly labels, and non-process anomaly labels, thereby completing the anomaly type classification task. Compared with the method of relying on human experience to perform secondary analysis of abnormal areas, this scheme completes the anomaly cause differentiation and labeling processing at the identification stage, so that the detection not only outputs the state results, but also forms a traceable anomaly source information chain, achieving the purpose of collaborative analysis of dyeing and printing process and structural bearing state. Attached Figure Description

[0053] Figure 1 This is a schematic diagram of the steps of a textile printing and dyeing identification method in the present invention.

[0054] Figure 2 This is a block diagram illustrating the logical principle of a textile dyeing and printing identification method in the present invention. Detailed Implementation

[0055] The technical solutions of the embodiments of the present invention will be clearly and completely characterized below with reference to the accompanying drawings. Obviously, the characterized embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0056] Example 1

[0057] Please see Figure 1 This invention provides a method for identifying printing and dyeing processes in textile printing and dyeing. To achieve the above objectives, this invention is implemented through the following technical solution, including the following steps:

[0058] S1. After the dyeing and printing are completed, the textile images are acquired by an industrial area array camera. The target area of ​​the textile image is located according to the target area positioning rules, and a set of local dyeing performance and a set of structural stability performance are constructed.

[0059] S2. Based on the local staining feature set, the image segment region is divided into several sub-regions within the local analysis window, and the color distribution structure features and color transition trend features are extracted and correlated to form staining response state features.

[0060] S3. Within the local analysis window, based on the set of structural stability performance, the image segment region is scanned pixel by pixel according to the row and column order of each sub-region to extract texture direction features and structural morphology features. After spatial alignment in the same local analysis window coordinate system, the structural bearing features are evaluated.

[0061] S4. Construct a set of staining-structure features and perform staining-structure coupling identification. Divide the state of the detection area based on the identification results.

[0062] S5. Based on the staining-structure coupling identification results, perform attribution analysis on the source of the anomaly and attach corresponding anomaly cause labels.

[0063] In this embodiment, steps S1 to S3 complete dual-path modeling of the image segment region: on the one hand, based on the local staining feature set, color distribution structure features and color transition trend features are extracted to form staining response state features; on the other hand, based on the structural stability performance set, the gray-level gradient direction is scanned pixel by pixel to extract texture direction features and structural morphology features, and structural bearing feature evaluation is completed. This process enables the staining performance and structural bearing state to be expressed synchronously in space, completing the hierarchical analysis of the printing and dyeing results from the color level to the structural level, rather than remaining at the level of judging a single color difference or overall texture statistics. Step S4 constructs a staining-structure feature set and performs staining-structure coupling identification, corresponding the staining response state features and structural bearing features to each region under the principle of spatial consistency, so that the division of the detection region state is based on the collaborative analysis of the dual feature system. Compared with the current detection method based on single-channel visual features, this method no longer outputs results solely based on color anomalies, but combines structural bearing features to perform spatial overlap relationship analysis, achieving a fine division of the detection region state. This mechanism completes the transformation from "whether it is abnormal" to "how the anomaly is related to the structure", giving the identification results a clear structural correlation logic. S5 performs attribution analysis on the sources of anomalies based on the dyeing-structure coupling identification results and attaches anomaly cause labels, so that each detection area not only has a status identifier but also a cause attribute. By distinguishing different types of anomaly sources, it achieves structured expression and categorized archiving of the identification results. Compared with technical means that rely on human experience for subsequent analysis, this method completes the anomaly source location and labeling at the identification stage, forming a complete closed-loop process from image acquisition, feature extraction, coupling identification to anomaly attribution, achieving the goal of collaborative analysis and fine classification of the dyeing process status and structural bearing status.

[0064] Example 2

[0065] Please refer to Figure 2 Specifically: S1 includes S11;

[0066] S11. Under the same working conditions after the dyeing and printing process is completed, the textile is imaged by an industrial area array camera set at the end of the dyeing and printing production line. The target area of ​​the textile image is located according to the target area positioning rules. After the positioning is completed, a local analysis window is constructed with the spatial coordinate range of the target area as the boundary, and the corresponding image segment area of ​​the target area is extracted. The position identifier and boundary information of the current local analysis window in the image are recorded, and the image segment area is output.

[0067] The target area includes the cuff edge area, the neckline splicing area, and the pattern edge area.

[0068] S1 also includes S12 and S13;

[0069] S12. Taking the image segment region as input, the original RGB data of the image is converted into CIELab color space data within the local analysis window. The color distribution density matrix is ​​calculated by performing partition histogram statistics and two-dimensional color co-occurrence matrix on the L, a, and b three-channel data. At the same time, the gradient of the color components of adjacent pixels is calculated by using the first-order difference operator to obtain the color transition gradient magnitude distribution. Then, the image color pixels are clustered based on the density clustering algorithm to extract the number of color clusters and the cluster area ratio. By performing variance and mean calculation on the pixel color values ​​within the local window, the color uniformity parameter is obtained, and a local coloring feature set is constructed.

[0070] S13. A grayscale image is generated by weighted summation of the RGB three-channel pixel values ​​of the image segment region. Gradient calculation is performed on the grayscale image to obtain the gradient direction and gradient magnitude of each pixel. The gradient direction data is statistically processed to count the frequency of occurrence and concentration of the main direction. Adjacent pixel pairs are set according to the preset pixel spacing and direction. The grayscale image is traversed pixel by pixel to count the occurrence of each grayscale level combination under the corresponding spacing and direction conditions. A grayscale level joint distribution table is constructed, and the joint distribution table is normalized to form a grayscale co-occurrence matrix. Based on the grayscale co-occurrence matrix, the concentration of grayscale difference distribution and the proportion of grayscale occurrence at the same level are calculated to obtain the texture grayscale change amplitude and texture arrangement consistency in the local area, and a structurally stable representation set is constructed.

[0071] In this embodiment, an industrial area array camera installed at the end of the dyeing and printing production line acquires images of textiles. Based on target area positioning rules, the cuff edge area, neckline splicing area, and pattern edge area are located. After determining the spatial coordinate range of the target area, a local analysis window is constructed, capturing the corresponding image fragment area and recording its location and boundary information. Then, within this local analysis window, on the one hand, the RGB data of the image fragment area is converted to CIELab color space data. Through the statistical analysis of the partition histograms of the L, a, and b channels, the construction of a two-dimensional color co-occurrence matrix, the calculation of the color transition gradient amplitude distribution using a first-order difference operator, and density clustering, the number of color clusters, the cluster area ratio, and color uniformity parameters are extracted to form a local dyeing feature set. On the other hand, a grayscale image is generated by weighted summation of the RGB three channels. The grayscale gradient direction and gradient amplitude are calculated, the frequency of occurrence of the main direction and the direction concentration are statistically analyzed, and the concentration of grayscale difference distribution and the proportion of grayscale occurrences at the same level are calculated based on the grayscale joint distribution table and the grayscale co-occurrence matrix to form a structurally stable performance set. Through the above steps, quantitative modeling of dyeing performance and structural morphology is completed simultaneously under the same local analysis window coordinate system, so that the color distribution state of the dyeing result corresponds to the stable state of the fabric texture. This not only completes the task of fine identification of key areas, but also provides a unified data foundation for subsequent dyeing-structure coupling identification. Compared with single color difference detection or overall texture statistics, it forms a judgment basis in terms of regional positioning accuracy, feature expression completeness and anomaly analysis basis.

[0072] Example 3

[0073] Please refer to Figure 2 Specifically: S2 includes S21, S22 and S23;

[0074] S21. Based on the local coloring feature set, the image segment region is divided into several sub-regions according to the image coordinates within the local analysis window. The mean chromaticity, chromaticity dispersion, and mean gradient amplitude are statistically calculated in each sub-region. The sub-regions are then arranged according to their spatial order to form color distribution structure features.

[0075] S22. Statistically analyze the gradient direction of each pixel, calculate the number of pixels in each direction interval, take the direction with the highest frequency as the main change direction, and traverse pixel by pixel along the main change direction, record the chromaticity difference change between consecutive pixels, when the chromaticity difference does not exceed the preset chromaticity threshold, record it as a gradient segment, when the chromaticity difference exceeds the preset chromaticity threshold, record it as a sudden change point, and merge consecutive sudden change points to determine the boundary position and generate color transition trend features.

[0076] S23. According to the sub-region number, associate the color distribution structure features of each sub-region with its corresponding color transition trend features to form a coloring response state feature that includes the region location, color distribution value, main direction angle, gradient length and boundary coordinates.

[0077] In this embodiment, the image segment region is first divided into several sub-regions according to image coordinates based on the local coloring feature set. The mean chromaticity, chromaticity dispersion, and mean gradient magnitude are statistically analyzed in each sub-region and arranged in spatial order to form a color distribution structure feature with positional correlation. Then, the pixel gradient direction is partitioned and statistically analyzed to determine the main change direction. The continuous chromaticity difference is analyzed pixel by pixel along this direction, and the chromaticity change is divided into gradual transition segments and abrupt change points. The continuous abrupt change points are merged to determine the boundary position and generate a color transition trend feature. Finally, the color distribution structure feature and the color transition trend feature are correlated according to the sub-region number to form a coloring response state feature that includes the region position, color distribution value, main direction angle, gradual length, and boundary coordinates. Through the above steps, the staining information is expanded from a single chromaticity statistic to a combination of features including spatial structure and transition behavior, thus completing the simultaneous characterization of color distribution patterns and color boundary change patterns. Compared with methods that rely solely on overall color difference or simple threshold judgments, this implementation constructs an ordered feature chain at the sub-region scale, enabling staining anomalies to be clearly expressed in terms of directionality, continuity, and boundary position, laying a data foundation for subsequent spatial alignment and coupling identification with structural bearing features.

[0078] Example 4

[0079] Please refer to Figure 2 Specifically: S3 includes S31 and S32;

[0080] S31. Within the local analysis window, based on the set of structurally stable performance, the image segment region is scanned pixel by pixel according to the row and column order of each sub-region. The gradient direction difference and gradient magnitude difference between adjacent pixels are counted. When the gradient direction difference between adjacent regions is within a preset range, it is determined to be a texture continuous region, and its spatial range is recorded. At the same time, the number of texture lines per unit area and their distribution difference are counted in each sub-region. The texture density is calculated based on the local gradient field aggregation degree to form the texture direction characteristics of the image segment region.

[0081] S32. On the grayscale image, the grayscale values ​​are continuously sampled along the main change direction to obtain a one-dimensional grayscale sequence. The grayscale sequence is subjected to differential processing to calculate the grayscale change between adjacent sampling points. The amplitude range of grayscale change and the interval between adjacent peaks are statistically calculated according to the statistical method to form surface undulation data and structural morphological features.

[0082] S3 also includes S33;

[0083] S33. Under the same local analysis window coordinate system, spatially align the boundary of the continuous texture region in the texture direction feature with the spacing offset position and boundary change position in the structural morphology feature, count the area ratio of the continuous texture region in the local analysis window, calculate the maximum value of the spacing offset value and the boundary position offset value and the length of the continuous over-limit segment in the continuous texture region, and then evaluate the structural bearing characteristics.

[0084] By dividing the statistical data of historical qualified samples and samples with structural anomalies under the same process conditions into intervals, the normal distribution range and abnormal distribution range of the corresponding parameters are calculated respectively, and the area threshold, spacing offset threshold and position offset threshold are set respectively at the boundary interval of the two types of sample distributions.

[0085] When the area ratio is not less than the preset area threshold, and the spacing offset value and the boundary position offset value do not exceed the corresponding spacing offset threshold and position offset threshold, output a structural stability status indicator.

[0086] When the area ratio is less than the preset area threshold, or the spacing offset value exceeds the corresponding spacing offset threshold, an abnormal structural status indicator is output.

[0087] When the area ratio is less than the preset area threshold and the boundary position offset value exceeds the position offset threshold, output a structural failure status indicator.

[0088] In this embodiment, firstly, the gray-level gradient direction is scanned pixel by pixel in the sub-region row and column order within the local analysis window. The gradient direction difference and gradient magnitude difference between adjacent pixels are statistically analyzed to identify continuous texture regions. The texture density is calculated by combining the number of texture lines per unit area and the local gradient field aggregation degree, forming a texture direction feature with spatial range identification. Secondly, gray values ​​are continuously sampled along the main change direction to construct a one-dimensional gray-level sequence. Surface undulation data is obtained through differential processing and peak interval statistics to form structural morphology features. On this basis, the spatial alignment of the boundary of the continuous texture region with the spacing offset position and the boundary change position is completed under the same local analysis window coordinate system. The area ratio, spacing offset value, boundary position offset value and continuous over-limit segment length are statistically analyzed. The area threshold, spacing offset threshold and position offset threshold are set by combining the interval division results of historical qualified samples and structurally abnormal samples, and the structural stability status identifier, structurally abnormal status identifier or structural failure status identifier is output. Through the above steps, a hierarchical evaluation of the structural carrying characteristics of image fragment regions is completed, enabling texture direction features and structural morphology features to form quantifiable criteria at the same spatial scale. Compared with the method of judging based solely on a single grayscale change or local texture statistics, this implementation incorporates continuity, density distribution, and morphological shift into a unified evaluation logic, realizing hierarchical recognition of structural stability and providing a structural state basis for subsequent staining-structure coupling identification.

[0089] Example 5

[0090] Please refer to Figure 2 Specifically: S4 includes S41 and S42;

[0091] S41. Based on the principle of spatial consistency, the dyeing response state characteristics and structural bearing characteristics are mapped region by region, and the dyeing response state characteristics and structural bearing characteristics after region-by-region mapping are unified to the same identification scale through feature comparison analysis technology to form a dyeing-structure feature set.

[0092] S42. Based on the coloring-structure feature set, perform coloring-structure coupling identification, and the specific identification is as follows;

[0093] When the area of ​​abnormal dyeing does not overlap with the area of ​​structural bearing characteristics in space, it is determined that the abnormal dyeing is caused by dyeing process factors, indicating that the abnormal dyeing is the dominant state.

[0094] When the coloring anomaly area and the structural bearing characteristic area overlap in space, it is determined that the coloring anomaly is caused by the structural bearing condition limitation, indicating that the structural bearing anomaly is the dominant state.

[0095] When both the dyeing response state and the structural load-bearing state are stable within the overall space, the dyeing process is considered to be in a state of structural and technological matching, indicating a state of structural-dyeing consistency.

[0096] S4 also includes S43;

[0097] S43. Analyze the staining-structure coupling identification results, convert the identification results into structured state information, and divide the current detection area state as follows;

[0098] When the dyeing abnormality is the dominant state, it means that the structural load-bearing state is within the normal range, but the dyeing performance is abnormally deviated. The abnormality comes from the dyeing process. At this time, the dyeing process adjustment suggestion is output and the relevant personnel are notified to make process corrections and improvements to the current area.

[0099] When structural load-bearing abnormalities dominate, it indicates that the dyeing is abnormal and the structural load-bearing capacity is insufficient. Continuing the dyeing process poses a quality risk. At this time, we will provide suggestions for adjusting the fabric tension and process the fabric condition before proceeding to the dyeing process.

[0100] When the structure-staining consistency is achieved, it indicates that there is consistency between the staining performance and the structural bearing condition, and no abnormal features are found. At this time, the current image segment region is judged to be in a qualified state and marked as a normal production sample.

[0101] In this embodiment, the dyeing response state features and structural bearing features are mapped region by region within the same local analysis window coordinate system. A one-to-one mapping relationship is established based on the principle of spatial consistency. Then, feature comparison analysis technology is used to unify the two types of features to the same identification scale, forming a dyeing-structure feature set. On this basis, dyeing-structure coupling identification is performed. By judging whether there is a spatial overlap between the dyeing abnormal area and the structural bearing feature area, the state of the detection area is distinguished, clearly divided into a dyeing abnormality-dominated state, a structural bearing abnormality-dominated state, or a structure-dyeing consistent state. Furthermore, the identification results are parsed into structured state information, bound to specific image segment regions, and output corresponding dyeing process adjustment suggestions or fabric tension adjustment suggestions, or marked as normal production samples. Through this implementation method, the dyeing response state and structural bearing state are jointly identified at the same spatial scale, completing the task of distinguishing and attributing the source of the anomaly. Compared with the technical means of judging only based on a single color feature, a recognition mechanism with spatial correspondence and cause-oriented logic is formed, making the state division of the detection area clearer and the anomaly handling path clearer. In this way, the dyeing identification result is transformed from a simple anomaly indication into a decision basis with process orientation.

[0102] Example 6

[0103] Please refer to Figure 2 Specifically: S5 includes S51;

[0104] S51. Based on the identification and transformation results, the current image segment region is identified by its region type, and the current image segment region is bound to the corresponding image segment region through the region positioning information to form a region category identifier.

[0105] The staining-structure coupling identification results are categorized based on the region category identifier. The identification results include the structure-staining consistency state, the staining abnormality-dominant state, and the structural bearing abnormality-dominant state, and the identification types are uniformly identified.

[0106] S5 also includes S52;

[0107] S52. Based on the identification type, perform anomaly source attribution analysis on the current identified object and attach corresponding anomaly cause labels. The anomaly cause labels include dyeing process-related anomaly labels, structure-related anomaly labels, and non-process anomaly labels.

[0108] When the identification results indicate that the dyeing response is abnormal but the structural load-bearing state is stable, the current abnormality will be marked as a dyeing process-related abnormality.

[0109] When the identification results indicate that the coloring anomaly is related to the structural bearing state, the current anomaly is marked as a structurally related anomaly.

[0110] When the identification results show that the anomaly is not persistent and does not conform to the process characteristics, the current anomaly is marked as a non-process anomaly.

[0111] In this embodiment, after completing the staining-structure coupling identification, the current image segment region is first identified by region type based on the identification transformation result. Combined with the region positioning information formed in the previous steps, the image segment region is bound to its spatial position in the overall image to form a region category identifier, so that each detection unit has a clear spatial affiliation and state category. Then, the structure-staining consistent state, staining abnormal dominant state, and structural load-bearing abnormal dominant state are uniformly classified and standardized based on the region category identifier. On this basis, the anomaly source attribution analysis in S52 is performed. According to the correspondence between the staining response state and the structural load-bearing state, staining process-related anomaly labels, structure-related anomaly labels, or non-process anomaly labels are added respectively, so that the identification result includes not only state judgment, but also cause attributes and category attributes. Through this implementation method, the analysis results of the staining response state features and structural bearing features formed in S1 to S4 are transformed into manageable and traceable regional category information and anomaly cause labels, realizing complete identification from image feature extraction, coupling identification to anomaly source location, and transforming the detection results from single state output to structured classification results, forming a systematic improvement in anomaly location accuracy, anomaly type differentiation clarity and regional level management capabilities.

[0112] Although embodiments of the invention have been shown and characterized, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended technical solutions and their equivalents.

Claims

1. A method for identifying printing and dyeing processes in textile printing and dyeing, characterized in that: Includes the following steps: S1. After the dyeing and printing are completed, the textile images are acquired by an industrial area array camera. The target area of ​​the textile image is located according to the target area positioning rules, and a set of local dyeing performance and a set of structural stability performance are constructed. S2. Based on the local staining feature set, the image segment region is divided into several sub-regions within the local analysis window, and the color distribution structure features and color transition trend features are extracted and correlated to form staining response state features. S3. Within the local analysis window, based on the set of structural stability performance, the image segment region is scanned pixel by pixel according to the row and column order of each sub-region to extract texture direction features and structural morphology features. After spatial alignment in the same local analysis window coordinate system, the structural bearing features are evaluated. S4. Construct a set of staining-structure features and perform staining-structure coupling identification. Divide the state of the detection area based on the identification results. S5. Based on the staining-structure coupling identification results, perform attribution analysis on the source of the anomaly and attach corresponding anomaly cause labels.

2. The dyeing and printing identification method in textile dyeing and printing process according to claim 1, characterized in that: S1 includes S11; S11. Under the same working conditions after the dyeing and printing process is completed, the textile is imaged by an industrial area array camera set at the end of the dyeing and printing production line. The target area of ​​the textile image is located according to the target area positioning rules. After the positioning is completed, a local analysis window is constructed with the spatial coordinate range of the target area as the boundary, and the corresponding image segment area of ​​the target area is extracted. The position identifier and boundary information of the current local analysis window in the image are recorded, and the image segment area is output. The target area includes the cuff edge area, the neckline splicing area, and the pattern edge area.

3. The dyeing and printing identification method in textile dyeing and printing process according to claim 2, characterized in that: S1 also includes S12 and S13; S12. Taking the image segment region as input, the original RGB data of the image is converted into CIELab color space data within the local analysis window. The color distribution density matrix is ​​calculated by performing partition histogram statistics and two-dimensional color co-occurrence matrix on the L, a, and b three-channel data. At the same time, the gradient of the color components of adjacent pixels is calculated by using the first-order difference operator to obtain the color transition gradient magnitude distribution. Then, the image color pixels are clustered based on the density clustering algorithm to extract the number of color clusters and the cluster area ratio. By performing variance and mean calculation on the pixel color values ​​within the local window, the color uniformity parameter is obtained, and a local coloring feature set is constructed. S13. A grayscale image is generated by weighted summation of the RGB three-channel pixel values ​​of the image segment region. Gradient calculation is performed on the grayscale image to obtain the gradient direction and gradient magnitude of each pixel. The gradient direction data is statistically processed to count the frequency of occurrence and concentration of the main direction. Adjacent pixel pairs are set according to the preset pixel spacing and direction. The grayscale image is traversed pixel by pixel to count the occurrence of each grayscale level combination under the corresponding spacing and direction conditions. A grayscale level joint distribution table is constructed, and the joint distribution table is normalized to form a grayscale co-occurrence matrix. Based on the grayscale co-occurrence matrix, the concentration of grayscale difference distribution and the proportion of grayscale occurrence at the same level are calculated to obtain the texture grayscale change amplitude and texture arrangement consistency in the local area, and a structurally stable representation set is constructed.

4. The dyeing and printing identification method in textile dyeing and printing process according to claim 3, characterized in that: S2 includes S21, S22 and S23; S21. Based on the local coloring feature set, the image segment region is divided into several sub-regions according to the image coordinates within the local analysis window. The mean chromaticity, chromaticity dispersion, and mean gradient amplitude are statistically calculated in each sub-region. The sub-regions are then arranged according to their spatial order to form color distribution structure features. S22. Statistically analyze the gradient direction of each pixel, calculate the number of pixels in each direction interval, take the direction with the highest frequency as the main change direction, and traverse pixel by pixel along the main change direction, record the chromaticity difference change between consecutive pixels, when the chromaticity difference does not exceed the preset chromaticity threshold, record it as a gradient segment, when the chromaticity difference exceeds the preset chromaticity threshold, record it as a sudden change point, and merge consecutive sudden change points to determine the boundary position and generate color transition trend features. S23. According to the sub-region number, associate the color distribution structure features of each sub-region with its corresponding color transition trend features to form a coloring response state feature that includes the region location, color distribution value, main direction angle, gradient length and boundary coordinates.

5. The dyeing and printing identification method in textile dyeing and printing process according to claim 3, characterized in that: S3 includes S31 and S32; S31. Within the local analysis window, based on the structural stability performance set, the gray-level gradient direction of the image segment region is scanned pixel by pixel according to the row and column order of each sub-region. The gradient direction difference and gradient magnitude difference between adjacent pixels are counted. When the gradient direction difference between adjacent regions is within a preset range, it is determined to be a texture continuous region, and its spatial range is recorded. At the same time, the number of texture lines per unit area and their distribution difference are counted in each sub-region. The texture density is calculated based on the local gradient field aggregation degree to form the texture direction characteristics of the image segment region. S32. On the grayscale image, the grayscale values ​​are continuously sampled along the main change direction to obtain a one-dimensional grayscale sequence. The grayscale sequence is subjected to differential processing to calculate the grayscale change between adjacent sampling points. The amplitude range of grayscale change and the interval between adjacent peaks are statistically calculated according to the statistical method to form surface undulation data and structural morphological features.

6. The dyeing and printing identification method in textile dyeing and printing process according to claim 5, characterized in that: S3 also includes S33; S33. Under the same local analysis window coordinate system, spatially align the boundary of the continuous texture region in the texture direction feature with the spacing offset position and boundary change position in the structural morphology feature, count the area ratio of the continuous texture region in the local analysis window, calculate the maximum value of the spacing offset value and the boundary position offset value and the length of the continuous over-limit segment in the continuous texture region, and then evaluate the structural bearing characteristics. By dividing the statistical data of historical qualified samples and samples with structural anomalies under the same process conditions into intervals, the normal distribution range and abnormal distribution range of the corresponding parameters are calculated respectively, and the area threshold, spacing offset threshold and position offset threshold are set respectively at the boundary interval of the two types of sample distributions. When the area ratio is not less than the preset area threshold, and the spacing offset value and the boundary position offset value do not exceed the corresponding spacing offset threshold and position offset threshold, output a structural stability status indicator. When the area ratio is less than the preset area threshold, or the spacing offset value exceeds the corresponding spacing offset threshold, an abnormal structural status indicator is output. When the area ratio is less than the preset area threshold and the boundary position offset value exceeds the position offset threshold, output a structural failure status indicator.

7. The dyeing and printing identification method in a textile dyeing and printing process according to claim 6, characterized in that: S4 includes S41 and S42; S41. Based on the principle of spatial consistency, the dyeing response state characteristics and structural bearing characteristics are mapped region by region, and the dyeing response state characteristics and structural bearing characteristics after region-by-region mapping are unified to the same identification scale through feature comparison analysis technology to form a dyeing-structure feature set. S42. Based on the coloring-structure feature set, perform coloring-structure coupling identification, and the specific identification is as follows; When the area of ​​abnormal dyeing does not overlap with the area of ​​structural bearing characteristics in space, it is determined that the abnormal dyeing is caused by dyeing process factors, indicating that the abnormal dyeing is the dominant state. When the coloring anomaly area and the structural bearing characteristic area overlap in space, it is determined that the coloring anomaly is caused by the structural bearing condition limitation, indicating that the structural bearing anomaly is the dominant state. When both the dyeing response state and the structural load-bearing state are stable within the overall space, the dyeing process is considered to be in a state of structural and technological matching, indicating a state of structural-dyeing consistency.

8. The dyeing and printing identification method in textile dyeing and printing process according to claim 7, characterized in that: S4 also includes S43; S43. Analyze the staining-structure coupling identification results, convert the identification results into structured state information, and divide the current detection area state as follows; When the dyeing abnormality is the dominant state, it means that the structural load-bearing state is within the normal range, but the dyeing performance is abnormally deviated. The abnormality comes from the dyeing process. At this time, the dyeing process adjustment suggestion is output and the relevant personnel are notified to make process corrections and improvements to the current area. When structural load-bearing abnormalities dominate, it indicates that the dyeing is abnormal and the structural load-bearing capacity is insufficient. Continuing the dyeing process poses a quality risk. At this time, we will provide suggestions for adjusting the fabric tension and process the fabric condition before proceeding to the dyeing process. When the structure-staining consistency is achieved, it indicates that there is consistency between the staining performance and the structural bearing condition, and no abnormal features are found. At this time, the current image segment region is judged to be in a qualified state and marked as a normal production sample.

9. The dyeing and printing identification method in a textile dyeing and printing process according to claim 8, characterized in that: S5 includes S51; S51. Based on the identification and transformation results, the current image segment region is identified by its region type, and the current image segment region is bound to the corresponding image segment region through the region positioning information to form a region category identifier. The staining-structure coupling identification results are categorized based on the region category identifier. The identification results include the structure-staining consistency state, the staining abnormality-dominant state, and the structure bearing abnormality-dominant state, and the identification types are uniformly identified.

10. The dyeing and printing identification method in a textile dyeing and printing process according to claim 9, characterized in that: S5 also includes S52; S52. Based on the identification type, perform anomaly source attribution analysis on the current identified object and attach corresponding anomaly cause labels. The anomaly cause labels include dyeing process-related anomaly labels, structure-related anomaly labels, and non-process anomaly labels. When the identification results indicate that the dyeing response is abnormal but the structural load-bearing state is stable, the current abnormality will be marked as a dyeing process-related abnormality. When the identification results indicate that the coloring anomaly is related to the structural bearing state, the current anomaly is marked as a structurally related anomaly. When the identification results show that the anomaly is not persistent and does not conform to the process characteristics, the current anomaly is marked as a non-process anomaly.

Images