Processing control method and system for suede super-imitation leather fabric

Abstract

The application relates to the technical field of image processing, and discloses a processing control method and system for suede super imitation leather fabric. The method comprises the following steps: collecting a suede fabric image through double-light-source illumination and preprocessing; extracting fabric texture features based on a Gabor filter and a local binary pattern; obtaining defect information by using a U-Net model analysis; calculating process parameters by using a genetic algorithm and double-face nap parameter collaborative control; and performing quality evaluation according to indexes such as density distribution and direction consistency. The application establishes a multidimensional online detection and intelligent control system, realizes accurate monitoring and self-adaptive adjustment of the fabric processing process, and thus overcomes the technical defects of single detection means, low control precision and response lag in the prior art.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of image processing, and particularly relates to a processing control method and system for suede super imitation leather fabric. BACKGROUND

[0002] In the existing processing technology of suede super imitation leather fabric, the process of combining bidirectional raising and air-jet weaving is generally adopted. The front surface of the fabric is treated by raising with 600-mesh diamond sand, the back surface is treated by raising with 400-mesh diamond sand, and the pile is directionally combed by using an air-jet device. In terms of quality detection, it mainly relies on manual experience and traditional physical test methods, such as hand feeling test, friction test and wear resistance test, etc. Some production lines use simple photoelectric detection devices to measure the raising effect by measuring the reflected light intensity of the fabric surface, or use a single image acquisition device to detect the appearance of the pile.

[0003] However, these traditional processing control methods have obvious deficiencies. First, the manual detection method is highly subjective and it is difficult to ensure the consistency of the detection standard; second, single photoelectric detection or image detection cannot fully reflect the quality status of the fabric, especially in the bidirectional raising process, it is difficult to simultaneously control the quality of the front and back surfaces; third, traditional detection methods are often offline detection, which cannot realize real-time monitoring and timely adjustment, and when quality problems are found, a large amount of unqualified products have already been produced; finally, the existing control methods lack intelligence and self-adaptive ability, and cannot automatically optimize the process parameters according to the detection results, resulting in large fluctuations in product quality and low production efficiency. SUMMARY

[0004] The present application provides a processing control method and system for suede super imitation leather fabric, which is used to realize accurate monitoring and self-adaptive adjustment of the fabric processing process, thereby overcoming the technical defects of single detection means, low control precision and lagging response in the prior art.

[0005] In a first aspect, the application provides a processing control method for suede super imitation leather fabric, which comprises: collecting images of the suede fabric through a double light source alternate illumination system to obtain complementary image data, and pre-processing the complementary image data to obtain enhanced fabric images; performing multi-dimensional feature extraction on the enhanced fabric images based on a Gabor filter bank and an improved local binary pattern to obtain fabric texture feature data; analyzing the fabric texture feature data according to a U-Net deep learning model to obtain fabric defect information; calculating process parameters based on a genetic algorithm and a double-sided nap parameter collaborative control model to obtain process parameter optimization data; and performing quality evaluation on the process parameter optimization data through pile density distribution, direction consistency, nap uniformity, fiber softness, surface glossiness and pile height difference to obtain fabric processing quality evaluation results.

[0006] In a second aspect, the application provides a processing control system for suede super imitation leather fabric, which comprises:

[0007] A collecting module is configured to collect images of the suede fabric through a double light source alternate illumination system to obtain complementary image data, and pre-process the complementary image data to obtain enhanced fabric images;

[0008] An extracting module is configured to perform multi-dimensional feature extraction on the enhanced fabric images based on a Gabor filter bank and an improved local binary pattern to obtain fabric texture feature data;

[0009] An analyzing module is configured to analyze the fabric texture feature data according to a U-Net deep learning model to obtain fabric defect information;

[0010] A calculating module is configured to calculate process parameters based on a genetic algorithm and a double-sided nap parameter collaborative control model to obtain process parameter optimization data;

[0011] An evaluating module is configured to perform quality evaluation on the process parameter optimization data through pile density distribution, direction consistency, nap uniformity, fiber softness, surface glossiness and pile height difference to obtain fabric processing quality evaluation results.

[0012] In a third aspect, the application provides a computer device, which comprises a memory and at least one processor, wherein the memory stores instructions; and the at least one processor invokes the instructions in the memory to enable the computer device to perform the processing control method for suede super imitation leather fabric.

[0013] The fourth aspect of the present application provides a computer-readable storage medium, wherein the computer-readable storage medium stores instructions which, when executed on a computer, cause the computer to perform the above-mentioned processing control method for suede super-imitation leather fabric.

[0014] In the technical solution provided in the present application, the suede fabric is imaged by a double-light-source alternating illumination system, so that the orthographic light source can accurately capture the surface texture features, and the oblique light source highlights the three-dimensional structure of the pile, the complementary image data generated by the alternately triggered two groups of light sources complement each other, effectively solving the problem that a single light source cannot fully reflect the surface features of the fabric. Multi-dimensional feature extraction is performed using a Gabor filter bank and an improved local binary pattern, wherein the Gabor filter bank can fully capture the pile direction features at different scales through six scale and eight direction parameter combinations, and the improved local binary pattern extracts rotation-invariant texture features using a circular neighborhood sampling strategy, and the combination of the two algorithms realizes accurate representation of the fabric texture features. The defect analysis method based on the U-Net deep learning model extracts multi-level features by using the ResNet50 backbone network in the encoder, introduces an attention mechanism in the decoder to enhance the perception ability of the defect area, and adds a directional convolution module in the skip connection, which significantly improves the detection sensitivity of the pile direction defects. The genetic algorithm and the double-sided pile parameter collaborative control model are used for process parameter optimization, the mapping relationship between the fabric defect features and the process parameters is established, and adaptive adjustment of the parameters is realized, and the introduction of the genetic algorithm ensures that the optimal solution is found in the complex parameter space. Finally, a comprehensive quality evaluation system is established through comprehensive evaluation of multiple dimensions such as pile density distribution, direction consistency, pile uniformity, fiber softness, surface glossiness and pile height difference, and accurate quantification of the processing quality of the fabric is realized. This intelligent control scheme based on artificial intelligence algorithm not only improves the accuracy and comprehensiveness of the detection, but also realizes closed-loop control of the processing process through real-time feedback and adaptive adjustment mechanism, significantly improving the processing quality and production efficiency of the suede super-imitation leather fabric. BRIEF DESCRIPTION OF DRAWINGS

[0015] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings needed in the embodiment description will be briefly introduced. Obviously, the drawings in the following description are some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0016] Figure 1 An embodiment schematic diagram of the processing control method for suede super-imitation leather fabric in the present application;

[0017] Figure 2A flowchart of a multi-dimensional feature extraction process on the enhanced fabric image in the embodiment of the present application is shown in the figure.

[0018] Figure 3 A schematic diagram of a matching relationship graph in the embodiment of the present application is shown in the figure.

[0019] Figure 4 An embodiment of a processing control system for suede super-imitation leather fabric in the embodiment of the present application is shown in the figure.

[0020] Figure 5 A structural schematic block diagram of a computer device in the embodiment of the present application is shown in the figure. DETAILED DESCRIPTION

[0021] The embodiment of the present application provides a processing control method and system for suede super-imitation leather fabric. The terms "first", "second", "third", "fourth" and the like (if any) in the specification and claims of the present application and the above-mentioned figures are used to distinguish similar objects, and do not necessarily mean a specific order or sequence. It should be understood that the data thus used can be interchanged under appropriate circumstances, so that the embodiments described herein can be implemented in an order other than that illustrated or described herein. In addition, the term "comprising" or "having" and any variations thereof are intended to cover non-exclusive inclusion, for example, a process, method, system, product or device including a series of steps or units does not necessarily have to be limited to those steps or units clearly listed, but can include other steps or units not clearly listed or inherent to these processes, methods, products or devices.

[0022] For the sake of understanding, the specific process of the embodiment of the present application is described below. Please refer to Figure 1 An embodiment of the processing control method for suede super-imitation leather fabric in the embodiment of the present application includes the following steps.

[0023] Step S101, image acquisition of suede fabric is performed by a double-light-source alternating illumination system to obtain complementary image data, and the complementary image data is preprocessed to obtain an enhanced fabric image.

[0024] Step S102, multi-dimensional feature extraction is performed on the enhanced fabric image based on a Gabor filter bank and an improved local binary pattern to obtain fabric texture feature data.

[0025] Step S103, fabric defect information is obtained by analyzing the fabric texture feature data according to a U-Net deep learning model.

[0026] Step S104, process parameter calculation is performed on the fabric defect information based on a genetic algorithm and a double-face nap parameter collaborative control model to obtain process parameter optimization data.

[0027] In step S105, the process parameter optimization data is quality evaluated by the pile density distribution, the direction consistency, the pile uniformity, the fiber softness, the surface glossiness and the pile height difference, and a fabric processing quality evaluation result is obtained.

[0028] It can be understood that the execution subject of the present application can be a processing control system for suede super-imitation leather fabric, and can also be a terminal or a server, which is not limited here. The server is taken as an example for description of the execution subject of the present application.

[0029] Specifically, the suede fabric is imaged by a 1200-megapixel high-speed industrial camera. The camera is fixed 2 meters above the production line with a field of view angle of 45°, and is matched with a LED ring light source with a color temperature of 5500K. When the fabric moves at a constant speed of 2m / min, the front light source in the double light source alternating illumination system captures the surface texture, and the side light source highlights the stereoscopic sense of the pile. The alternating triggering of the two groups of light sources is realized by a time sequence controller, and the complementary image data with a resolution of 4096x3072 pixels and a bit depth of 12 bits are generated respectively. The acquired complementary image data is preprocessed, and the CLAHE algorithm is used for contrast enhancement. The image is divided into 64x64 grids, and each grid is histogram equalized with a clip limit of 3.0. Then, the improved anisotropic diffusion filter is used for noise reduction, and the diffusion coefficient is adaptively adjusted according to the local gradient. For the pile area, a directional enhancement operator based on morphology is designed, and the pile texture is extracted using 8-direction structural elements. The Retinex algorithm is used for non-uniform illumination correction, and the enhanced fabric image is obtained by dynamically optimizing the algorithm parameters according to the color characteristics of the fabric.

[0030] Based on the obtained enhanced fabric image, a Gabor filter bank is constructed for multi-dimensional feature extraction. The filter parameters include 6 scales and 8 directions, and the pile direction features are extracted. Multi-scale texture information is obtained through wavelet packet decomposition, and the sym4 wavelet basis is selected with 4 levels of decomposition. A new type of local binary pattern operator is designed, with a circular neighborhood radius R = 2 and a sampling point number P = 16, to extract rotation-invariant texture features. The gray level co-occurrence matrix features are calculated, including energy, entropy, contrast, and other statistical quantities. Finally, principal component analysis is used to reduce the dimension of the features, retaining 95% of the information, to generate the fabric texture feature data. An improved U-Net structure is used for fabric texture feature data defect detection analysis. The network encoder uses a ResNet50 backbone network, and the decoder incorporates an attention mechanism. Directional convolution modules are added in the skip connection to enhance the sensitivity to pile direction defects. The loss function combines pixel-level cross-entropy and boundary IoU loss to improve the accuracy of defect boundary positioning. The model training uses a transfer learning strategy, first pre-training on a general textile dataset, then fine-tuning on the suede sample, and finally outputting the fabric defect information.

[0031] A defect-process parameter correlation matrix is constructed, including key parameters such as pile height, pile density, and fabric tension. Genetic algorithm and fuzzy inference system are used for process parameter optimization, with the input variable being the defect feature vector and the output variable being the process parameter adjustment. The fuzzy rule base is established based on expert experience and historical data, and the membership function parameters are optimized through genetic algorithm. Reinforcement learning mechanism is introduced to continuously update the decision strategy according to the adjustment effect, and process parameter optimization data is generated. The process parameter optimization data is evaluated by a multi-dimensional quality evaluation system. Pile density distribution, pile direction consistency, pile uniformity, fiber softness, surface glossiness, and pile height difference are calculated. Pile morphology features including length distribution, density distribution, and direction distribution are calculated through image processing algorithms. A hand feeling prediction model is constructed based on multi-sensor data for comprehensive quality evaluation, and the fabric processing quality evaluation results are output.

[0032] Taking the production of a batch of suede fabric as an example, when the collected orthographic light source image shows that the density distribution of the fabric surface pile is uneven, the deviation of the three-dimensional structure of the pile is further confirmed through the complementary side light source image. After CLAHE algorithm processing, the local contrast is improved, and the pile texture is clearer. Gabor filter analysis finds that the main direction deviates by 15 degrees, and the improved local binary pattern shows that the texture consistency is reduced. The U-Net model detects multiple small defect points distributed in the edge area of the fabric. Based on these defect features, the genetic algorithm suggests adjusting the front nap speed from 2000 rpm to 1900 rpm and the grit mesh from 600 mesh to 550 mesh. After parameter adjustment, quality evaluation shows that the uniformity of pile density distribution is improved, the direction consistency is corrected from the original deviation to the standard range, and the surface gloss reaches the expected level.

[0033] In the embodiments of the present application,

[0034] In a specific embodiment, the process of step S101 can specifically include the following steps:

[0035] (1) Collect the fabric motion image sequence through a high-speed industrial camera, and combine the motion trajectory compensation algorithm to eliminate the motion blur of the image sequence to generate the original image;

[0036] (2) The surface texture features of the orthographic light source data and the three-dimensional contour information of the side light source data are fused in time sequence through an adaptive weight fusion algorithm to obtain double light source complementary image data;

[0037] (3) Based on the image local entropy value and gradient distribution, the double light source complementary image data is adaptively divided into grids, and the grid size is dynamically determined by the texture complexity to obtain segmented grid data;

[0038] (4) The segmented grid data is histogram equalized by using local contrast mapping and global brightness calibration, and the best stretching coefficient is determined through iterative optimization to obtain an equalized image;

[0039] (5) According to the pile direction field distribution, a multi-scale and multi-direction morphological operator is constructed to enhance the pile texture of the equalized image, and edge preserving filtering is used to suppress noise to obtain texture enhanced data;

[0040] (6) Based on the illumination compensation model of the fabric color features, the texture enhanced data is nonlinearly corrected and contrast optimized, and the parameters are adaptively adjusted according to the illumination non-uniformity evaluation index to obtain the enhanced fabric image.

[0041] Specifically, when using a high-speed industrial camera to collect images of the fabric on the production line, the fabric moves at a constant speed of 2 m / min, which causes motion blur of the images. To this end, a motion blur metric function is designed:

[0042]

[0043] wherein, represents the gray value of the blurred image at point (x, y), I(x, y) is an ideal clear image, is a blur coefficient, and represents the displacement deviation of the i-th sampling point, and N is the number of sampling points. When the fabric moves at a speed of 2 m / min, the displacement of the fabric during each camera exposure is about 0.2 mm, which corresponds to a shift of 3-5 pixels in the image space. The blur caused by the motion can be determined by the blur metric function, and then targeted compensation can be performed. After obtaining the original image, the data collected by the orthographic and oblique light sources need to be fused. The orthographic light source is mainly used to collect surface texture features, and the oblique light source highlights the three-dimensional profile information of the fabric. An adaptive weight fusion algorithm is used:

[0044]

[0045] wherein, is the value of the fused pixel point p, is the surface texture feature value, is the three-dimensional profile information value, and is an adaptive weight coefficient, and + = 1. In actual application, when it is detected that the pile density of a certain area of the fabric is high, the value of is increased to highlight the texture features; when the area has sparse pile, the value of is increased to enhance the three-dimensional information. For example, in the dense pile area, can reach 0.7-0.8, while in the sparse area, it decreases to 0.3-0.4.

[0046] In order to enhance the performance of the pile texture, a multi-scale morphological enhancement operator is constructed:

[0047]

[0048] wherein, is the enhanced texture intensity, is the k-th scale image, is a structure element, and represent dilation and erosion operations, respectively, Here, M represents the weighting coefficients for each scale, and M is the number of scales. When processing suede fabric images, 3-4 scales are typically used, with structuring element sizes ranging from 3×3 to 9×9 pixels. When the fabric surface nap is detected to be arranged at a 45-degree angle, the structuring element is rotated by 45 degrees accordingly to make the enhancement effect more targeted. Weighting coefficients The algorithm gradually decreases the size of the image as the scale increases, ensuring that detailed textures are sufficiently enhanced without overemphasizing large-scale features. A complete workflow from image acquisition to preprocessing is implemented. For example, when processing a batch of suede fabric, a 5-pixel motion offset was determined using motion blur measurement, and after compensation and restoration, a clear original image was obtained. Subsequently, in the dual-light source fusion stage, it was found that the pile was denser on the left side of the fabric. The algorithm automatically adjusted the texture weight of this area to 0.75, while reducing it to 0.35 in the sparser area on the right side. Finally, multi-scale morphological enhancement was used, employing structuring elements of three scales: 3×3, 5×5, and 7×7, with weights of 0.5, 0.3, and 0.2 respectively, successfully enhancing the layering of the pile texture.

[0049] In one specific embodiment, the process of performing step S102 may specifically include the following steps:

[0050] (1) Construct a six-scale, eight-direction Gabor filter bank for the enhanced fabric image to perform directional feature analysis and obtain the directional feature data of the pile;

[0051] (2) Based on the sym4 wavelet basis, the directional feature data of the fluff is decomposed into four levels, and the multi-scale texture coefficients of double-sided napping of 600 mesh and 400 mesh are extracted respectively.

[0052] (3) Binarize the multi-scale texture coefficients by sampling in a circular neighborhood to generate rotation-invariant feature sequences of the front and back surfaces;

[0053] (4) Calculate the energy, entropy, and contrast feature values ​​for the rotation-invariant feature sequence, and construct the gray-scale statistical matrix of the double-sided texture;

[0054] (5) Perform humidity compensation correction on the gray-scale statistical matrix based on the 25-30% humidity parameter to obtain the compensated feature vector;

[0055] (6) The compensated feature vector is weighted and fused with the double-sided pile direction feature data to generate fabric texture feature data.

[0056] Specifically, such as Figure 2 The diagram shown illustrates the process of multi-dimensional feature extraction from the enhanced fabric image in this embodiment of the application. In the processing control of suede-like super-imitation leather fabric, a Gabor filter bank is constructed to perform directional feature analysis on the fabric image. The mathematical expression for the Gabor filter is:

[0057]

[0058] Where r represents the distance from the pixel to the center (in pixels). Indicates direction angle (unit: radians), and The standard deviations are the radial and angular directions, respectively. The center frequency is used. In practical applications, six scales are used in the processing of a batch of suede fabric. (2, 4, 6, 8, 10, 12 pixels respectively) and eight directions ( The filter banks (0°, 45°, 90°, 135°, 180°, 225°, 270°, 315°) are used to extract the directional features of the 600-mesh front fuzz and the 400-mesh back fuzz.

[0059] Wavelet transform analysis was performed on the obtained fuzz orientation feature data. The formula for calculating wavelet coefficients was constructed as follows:

[0060]

[0061] in, Let be the wavelet coefficients in the p direction of the j-th layer. Let sym4 be the wavelet basis function, D be the input directional feature data, and (m,n) be the coordinate position. During the processing of suede fabric, a four-level decomposition is performed to obtain high-frequency and low-frequency sub-band coefficients. When a directional deviation in the fabric surface nap is detected, the location and range of the abnormal area are determined by analyzing the amplitude distribution of the wavelet coefficients at each level.

[0062] Finally, statistical analysis of texture features is performed, and the formula for calculating the gray-level co-occurrence matrix is ​​introduced:

[0063]

[0064] in, Let the distance be d and the direction be... Let I be the gray-level co-occurrence matrix, where I is the input image and (i,k) are gray-level value pairs. In practical applications, d = 1, 2, 3 pixels is selected. The system calculates characteristic values ​​such as energy, entropy, and contrast in four directions: 0°, 45°, 90°, and 135°. For fabrics exposed to 25-30% humidity, the characteristic values ​​need to be corrected according to the humidity coefficient. For example, in processing a batch of fabric, when the ambient humidity is 28%, the calculated characteristic values ​​are multiplied by a compensation coefficient of 1.15 to eliminate the influence of humidity on texture characteristics.

[0065] Through the combination of the three key algorithms, the complete analysis process from direction feature extraction to texture feature statistics is completed. In the processing of suede fabric in actual production, the Gabor filter is used to detect that the direction of the pile is 75°, and the wavelet decomposition finds that there are abnormal coefficients in the third layer high-frequency subband, indicating that there is a local direction deviation in this position. Finally, through the gray level co-occurrence matrix analysis, it is determined that the energy value of this area is low and the contrast is high, and it is judged that the pile process parameters need to be adjusted.

[0066] In a specific embodiment, the process of performing step S103 can specifically include the following steps:

[0067] (1) The fabric texture feature data is subjected to feature hierarchical extraction through a multi-layer convolution progressive network to obtain a hierarchical feature mapping;

[0068] (2) The hierarchical feature mapping is subjected to double weighting operation based on channel attention mechanism and spatial attention mechanism to obtain an enhanced feature map;

[0069] (3) The enhanced feature map is subjected to pile texture analysis through a multi-directional convolution kernel group to generate front and back direction defect features;

[0070] (4) The front and back direction defect features are subjected to pixel-level depth segmentation and boundary constraint detection to obtain double-sided defect data;

[0071] (5) The pressure coefficient and the steam parameter are used to correct the double-sided defect data to obtain a compensated defect matrix;

[0072] (6) The compensated defect matrix is subjected to defect classification in combination with defect morphological features and texture statistical features to generate fabric defect information.

[0073] Specifically, in the defect detection process of suede super-imitation leather fabric, multi-level analysis is performed on the fabric texture feature data. A feature extraction function is constructed through a multi-layer convolution progressive network:

[0074]

[0075] wherein, is the lth layer feature map, is the convolution kernel weight matrix, is the input feature, (x, y) is the feature map coordinate, and is the convolution kernel size. In the processing of suede fabric, a 5-layer convolution network is used, and the convolution kernel size increases from 3x3 to 11x11, and the feature information from detail to global is extracted layer by layer.

[0076] The obtained hierarchical feature mapping is enhanced by introducing attention mechanism. A double attention calculation formula is designed:

[0077]

[0078] wherein, is an attention weight, is a channel attention coefficient, is a spatial attention coefficient, is an original feature value, and are channel and spatial weight factors, respectively. In actual application, when a suspected defect is detected in a certain area of the fabric, the network pays more attention to the feature performance of the area by adjusting the attention weight. For example, when an abnormal texture is found in the 600-mesh area on the front side, the attention coefficient of the corresponding channel will automatically increase to more than 0.8.

[0079] Finally, the environmental compensation and classification of the defect features are performed. An environmental compensation function is introduced as follows:

[0080]

[0081] wherein, is a compensated defect feature, is a pressure coefficient matrix, is a temperature coefficient matrix, is an original defect feature, and (h, w) is a feature matrix coordinate, and are feature matrix dimensions. In actual production, when the fabric is under a 0.2-0.3 MPa layering pressure and a 40°C steam temperature environment, the detected defect features need to be corrected by the function. For example, in a batch of fabric detection, it is found that the 400-mesh area on the back side has uneven phenomenon, and after correction by the pressure compensation coefficient 1.2 and the temperature compensation coefficient 0.85, the false defect caused by environmental factors is accurately identified, avoiding misjudgment.

[0082] Through the cooperation of the three core algorithms, a complete analysis process from feature extraction to defect recognition is completed. In actual production line, the method successfully detects various types of fabric defects. For example, in a detection, the system finds abnormal features in the edge area through the multi-layer convolution network, and the attention mechanism immediately increases the weight of the area to 0.85. Further analysis finds that it is due to the uneven layering pressure that leads to local insufficient pilling. After environmental factor compensation and feature classification, the accurate defect type and position information are output.

[0083] In a specific embodiment, the process of performing step S104 can specifically include the following steps:

[0084] (1) The fabric defect information is combined with the pile height, pile density, and fabric tension parameter to construct a parameter association matrix, and an initial process parameter is generated;

[0085] (2) The initial process parameter is analyzed for front and back coordination, a matching relationship diagram of double-sided pile speed and diamond grit number is established, and coordinated parameter data is obtained;

[0086] (3) The coordinated parameter data is graded and mapped by a fuzzy membership function, and a process parameter adjustment sequence is obtained;

[0087] (4) The process parameter adjustment sequence is iteratively updated based on a reinforcement learning strategy, and an optimized parameter combination is obtained;

[0088] (5) The steam temperature and fabric humidity are used to compensate for the optimized parameter combination, and a compensated parameter matrix is generated;

[0089] (6) The compensated parameter matrix is balanced and verified for both sides, and process parameter optimization data is obtained.

[0090] Specifically, the association between fabric defect information and process parameters is established. The defect information includes location, type, size, and other characteristic data, while the process parameters mainly include pile height (unit: mm), pile density (root / mm²), fabric tension (N / cm), etc. Through data statistical analysis, an association matrix is constructed. Taking pile height as an example, when the defect type is uneven pile, the actual measured height value of the defect area is compared with the standard height value, the height deviation is calculated, and the adjustment amount of the pile height is determined. Similarly, the pile density parameter is determined by the difference between the measured density value of the defect area and the standard value. The fabric tension is adjusted according to the defect morphology characteristics. The combination of these parameters constitutes the initial process parameter.

[0091] The front and back coordination analysis stage focuses on the balance of double-sided pile effect. According to the initial process parameter, the matching relationship of front 600 grit and back 400 grit diamond pile is analyzed. The speed range is 1800-2000 rpm for the front and 1600-1800 rpm for the back. By establishing a corresponding relationship diagram of speed and number of eyes, the best matching point is determined. For example, as shown in FIG. 2, the front 600 grit and back 400 grit diamond pile speed matching relationship diagram is established, and the best matching point is determined as 1900 rpm for the front and 1700 rpm for the back. Figure 3As shown, the matching relationship diagram in the embodiment of the present application is a schematic diagram, when the front side adopts 600 mesh diamond sand and the rotation speed is 1900 rpm, the back side needs to adopt 400 mesh diamond sand and the rotation speed is 1700 rpm, so as to achieve the ideal double-side raising effect. This matching relationship is summarized through a large amount of experimental data to generate collaborative parameter data. When the collaborative parameter data is classified and mapped, a fuzzy membership function is used for processing. The process parameter value range is divided into multiple levels, such as five levels of “too low”, “slightly low”, “moderate”, “slightly high” and “too high” for raising height. Each parameter value has different degrees of membership to these levels. Through fuzzy rule reasoning, the direction and amplitude of parameter adjustment are obtained to form a process parameter adjustment sequence.

[0092] In the iterative optimization phase, the parameter adjustment sequence is updated using a reinforcement learning strategy. Set the parameter adjustment target, such as improving the uniformity of the pile and enhancing the consistency of the direction. After each parameter adjustment, evaluate the adjustment effect and adjust the strategy according to the effect feedback. Good adjustment effect obtains positive reward, otherwise negative reward. Through multiple iterations, the optimized parameter combination is obtained. The environmental compensation phase focuses on the influence of steam temperature and fabric moisture. The steam temperature is controlled at 40°C, and the fabric moisture is maintained within the range of 25-30%. When the environmental parameters fluctuate, the process parameters need to be compensated accordingly. For example, when the steam temperature decreases, the raising rotation speed needs to be appropriately increased; when the fabric moisture increases, the layering pressure needs to be adjusted. These compensation relationships are recorded in the compensation parameter matrix.

[0093] Finally, the double-side balance verification is performed to ensure the consistency of the raising effect of the front and back sides. By comparing the pile height, density and direction of the front and back sides, the balance index is calculated. When imbalance is found, the process parameters need to be adjusted again until the predetermined balance standard is reached to obtain the process parameter optimization data. In actual production, taking the processing of a batch of suede fabric as an example, it is found that there are sparse defects in the edge area of the fabric. Through correlation matrix analysis, it is concluded that the raising intensity of this area needs to be increased. Collaborative analysis shows that the front side raising rotation speed needs to be increased from 1900 rpm to 1950 rpm, and the back side rotation speed needs to be adjusted to 1750 rpm accordingly. Through fuzzy rule reasoning, a step-by-step adjustment strategy is determined: first increase the front side rotation speed by 25 rpm, and then gradually adjust after observing the effect. Through multiple iterations and optimization, the best effect is achieved at 1940 rpm on the front side and 1740 rpm on the back side. Considering that the steam temperature is 2°C lower and the fabric moisture is 28% at that time, the parameters are compensated for the environment. The double-side balance verification shows that the difference in pile height between the front and back sides is controlled within 0.1 mm, and the density uniformity meets the expected standard, successfully solving the problem of sparse pile in the edge area.

[0094] In a specific embodiment, the process of step S105 can specifically include the following steps:

[0095] (1) The fabric sample is segmented by process parameter optimization data to obtain evaluation unit data, and the evaluation unit data is statistically operated by unit area pile count to generate pile density distribution;

[0096] (2) The evaluation unit data is scanned on both sides, and the direction consistency is calculated according to the deviation of the pile orientation angle, and the uniformity of the evaluation unit data is operated by regional density variance analysis to obtain the uniformity of the pile;

[0097] (3) The evaluation unit data is measured by pressure response, and the fiber softness is calculated according to the pressure deformation curve;

[0098] (4) The evaluation unit data is measured by multi-angle optical scanning to generate surface gloss;

[0099] (5) The evaluation unit data is measured by double-side height, and the pile height difference is calculated according to the height difference between the front and back;

[0100] (6) The pile density distribution, direction consistency, pile uniformity, fiber softness, surface gloss and pile height difference are weighted to obtain the fabric processing quality evaluation result.

[0101] Specifically, the fabric is segmented according to the process parameter optimization data. The region segmentation adopts a standard evaluation unit of 10 cm x 10 cm, forming a grid-shaped evaluation region on the surface of the fabric. Each evaluation unit is image collected by a high-resolution camera, and the camera resolution is 12 million pixels, the collection distance is 2 meters, and the image definition is ensured. The collected evaluation unit data is counted by pile, and each pile is identified and marked by image processing technology, and the number of piles in unit area is calculated. The pile density distribution data is obtained by scanning and counting the piles in each evaluation unit. After obtaining the evaluation unit data, the front and back surfaces are scanned and analyzed simultaneously. A double-head scanning device is used to scan the pile direction of the front and back surfaces of the fabric. The scanning device contains a direction sensor that can detect the inclination angle of the pile. The front pile is raised by 600 mesh diamond sand, and the back pile is raised by 400 mesh diamond sand. The pile direction of the two surfaces should be consistent in theory. The direction consistency data is obtained by calculating the deviation value of the actual measurement angle and the theoretical angle. At the same time, the pile density in each evaluation unit is analyzed by variance, and the dispersion degree of the density in the region is calculated to obtain the pile uniformity data.

[0102] The pressure response measurement is an important means to evaluate the fiber softness of the fabric. A pressure testing device is used to apply a standard pressure to the evaluation unit, with a pressure range from 0 to 0.3 MPa, and the corresponding relationship between pressure and deformation is recorded. During the test, the pressure is increased step by step with a step size of 0.05 MPa, and each pressure point stays for 5 seconds, and the steady-state deformation value is recorded. By analyzing the slope change of the pressure-deformation curve, the softness index of the fabric is calculated. The surface glossiness measurement uses a multi-angle glossiness instrument to measure the reflection intensity at different angles of the evaluation unit. The measurement angles include 20°, 45°, 60°, and 75° standard angles, and the light source uses a standard D65 light source. The reflection intensity value is recorded at each angle to construct a reflection intensity curve. By analyzing the reflection intensity difference at different angles, the surface glossiness index is calculated.

[0103] The pile height difference measurement uses laser ranging technology to scan the height of the front and back of the fabric. The scanning resolution is 0.1 mm, and 100 measurement points are taken for each evaluation unit. By calculating the height difference value of the corresponding positions of the front and back, the height difference distribution map is obtained. Statistical analysis is performed on the height difference data to calculate the average difference value and the standard deviation, thereby evaluating the uniformity of the pile height. Finally, the comprehensive quality evaluation is performed, and the various indicators are weighted and calculated. The weight distribution is based on the pile density distribution as the basic indicator, the direction consistency and the pile uniformity as the important supplementary indicators, and the fiber softness, the surface glossiness and the pile height difference as the auxiliary evaluation indicators. After standardizing the data of each indicator, the weighted sum is calculated according to the preset weight coefficient to obtain the quality evaluation score.

[0104] Taking the production of a batch of suede fabric as an example, the fabric is divided into several 10 cm x 10 cm evaluation units during the quality evaluation process. Through pile counting, it is found that the pile density in the center area is relatively uniform, with about 15-18 piles per square millimeter, while the density in the edge area decreases slightly to 12-15 piles / mm². The front and back scanning shows that the main direction of the pile in the 600-mesh pile area on the front surface is 75°, and the main direction of the pile in the 400-mesh area on the back surface is 73°, with a direction deviation within the acceptable range. The pressure test records show that under a pressure of 0.2 MPa, the fabric compression deformation is 2.3 mm, and the rebound is good after release. In the multi-angle glossiness measurement, the reflection intensity is highest at an incident angle of 45°, which is close to the standard sample. The front and back height difference measurement shows that the height difference in the center area is within 0.2 mm, and the height difference in the edge area increases slightly. Based on the data of various indicators, the fabric processing quality evaluation result is generated, and it is pointed out that the pile strength in the edge area needs to be appropriately improved to improve the uniformity of the pile density.

[0105] In a specific embodiment, the process of performing step S106 can specifically include the following steps:

[0106] (1) The quality correction probe is used for data hierarchical scanning of the fabric processing quality evaluation results, and quality distribution surface data is generated;

[0107] (2) Time sequence comparison operation is performed on the quality distribution surface data, and a dynamic control sequence is generated according to the quality fluctuation curve calculation;

[0108] (3) The dynamic control sequence is analyzed by a multi-point sensing network to obtain a process deviation matrix;

[0109] (4) The process deviation matrix is subjected to partition operation to generate a deviation compensation curve, and the process parameters are dynamically configured according to the deviation compensation curve to obtain a parameter adjustment sequence;

[0110] (5) The process control link is constructed according to the parameter adjustment sequence, and the control instruction of the fabric processing equipment is decomposed to generate equipment control data flow;

[0111] (6) The equipment control data flow is used for cooperative scheduling of the process execution unit to obtain execution feedback data;

[0112] (7) The execution feedback data and the fabric processing quality evaluation results are subjected to closed-loop iterative calculation to generate quality optimization control instructions.

[0113] Specifically, the quality correction probe is a sensing device specially designed for accurate measurement of fabric quality parameters. The probe contains a high-precision photoelectric sensor array with a spacing of 1 mm covering the fabric width. When the probe performs hierarchical scanning on the fabric processing quality evaluation results, an automatic stepping method is adopted with a step length of 0.5 mm, and the scanning range covers the entire fabric surface. Each scanning point records the quality parameter value at the corresponding position, including pile density, direction consistency, and pile uniformity, etc. indicators, forming a three-dimensional data point cloud. Through data interpolation algorithm, these discrete points are connected into a continuous quality distribution surface. The quality distribution surface data is processed by time sequence analysis method. According to the time sequence of fabric production, the continuously collected quality distribution surface data is arranged into a time sequence. The quality data of adjacent time is calculated by difference every 1 minute to obtain the quality change rate. According to the positive and negative and size of the change rate, the quality fluctuation curve is drawn. When the fluctuation curve shows obvious trend change, the process parameters need to be adjusted in time. By analyzing the change rule of the fluctuation curve, the timing and amplitude of process parameter adjustment are determined to generate a dynamic control sequence.

[0114] The multi-point sensing network is composed of sensors distributed at key points of the fabric processing equipment, including temperature sensors, pressure sensors, displacement sensors, etc. These sensors collect real-time process parameter data with a sampling frequency of 10 Hz. The collected data are matched and analyzed with the dynamic control sequence to calculate the deviation between the actual process parameters and the control requirements, and a process deviation matrix is constructed. The deviation matrix reflects the parameter errors of each process link and provides a basis for parameter adjustment. The process deviation matrix is processed using a partition calculation method. The fabric is divided into different functional areas such as edge area and middle area, and different deviation tolerances are set for each area. A continuous deviation compensation curve is generated through spline interpolation algorithm, and the slope of the curve represents the rate of parameter adjustment. According to the compensation curve, a parameter adjustment strategy is designed to generate a parameter adjustment sequence containing adjustment time, adjustment amount, and adjustment sequence.

[0115] The process control link is a processing process that converts the parameter adjustment sequence into specific equipment control instructions. It includes speed adjustment of the raising machine, pressure control, temperature and humidity adjustment, and other execution links. The control instructions are decomposed according to the response characteristics of the equipment to ensure smooth response of each execution unit. The control data stream contains information such as instruction type, execution time, and target value. The execution unit includes drive motors, pneumatic actuators, heating devices, and other specific execution mechanisms. These mechanisms receive instructions from the control data stream and perform corresponding actions according to the predetermined sequence. After each execution action is completed, the relevant sensors collect execution effect data to generate execution feedback data. The feedback data includes actual execution parameter values and execution time information.

[0116] Finally, closed-loop iterative calculation is performed to compare the execution feedback data with the original quality evaluation results. The difference between the execution results and the target requirements is calculated to evaluate the effect of parameter adjustment. Based on the evaluation results, the control strategy is corrected to generate new quality optimization control instructions.

[0117] Taking a batch of suede fabric in actual production as an example, the quality correction probe found that the pile density in the middle area of the fabric fluctuated during the scanning process. The data layer scanning showed that the pile density in this area was lower than the standard value, and the quality distribution surface presented a concave shape. Time series analysis showed that this fluctuation was gradual and gradually appeared after 30 minutes of production. The multi-point sensing network data showed that the pressure of the raising machine fluctuated during this period, and the speed was basically stable. The process deviation matrix analysis determined that the raising pressure needed to be adjusted. The generated compensation curve indicated that the pressure should be gradually increased within 5 minutes. The parameter adjustment sequence was converted into control instructions, and the pneumatic actuator gradually increased the pressure. The execution feedback showed that after the pressure adjustment was completed, the pile density began to rise, and the fluctuation was effectively controlled.

[0118] The processing control method for suede super imitation leather fabric in the embodiment of the application is described above, and the processing control system for suede super imitation leather fabric in the embodiment of the application is described below. Please refer to Figure 4 An embodiment of the processing control system for suede super imitation leather fabric in the embodiment of the application comprises:

[0119] The acquisition module is configured to acquire images of the suede fabric through a double-light-source alternating illumination system to obtain complementary image data, and to pre-process the complementary image data to obtain enhanced fabric images.

[0120] The extraction module is configured to extract multi-dimensional features of the enhanced fabric images based on a Gabor filter bank and an improved local binary pattern to obtain fabric texture feature data.

[0121] The analysis module is configured to analyze the fabric texture feature data according to a U-Net deep learning model to obtain fabric defect information.

[0122] The calculation module is configured to calculate process parameters based on a genetic algorithm and a double-face nap parameter collaborative control model to obtain process parameter optimization data.

[0123] The evaluation module is configured to evaluate the process parameter optimization data through pile density distribution, direction consistency, nap uniformity, fiber softness, surface glossiness, and pile height difference to obtain fabric processing quality evaluation results.

[0124] Through the cooperation of the above-mentioned components, the image acquisition of suede fabric is carried out through the double-light-source alternate illumination system, so that the orthographic light source can accurately capture the surface texture features, and the oblique light source highlights the stereoscopic structure of the pile, the complementary image data generated by the alternately triggered two groups of light sources complement each other, effectively solving the problem that a single light source is difficult to fully reflect the surface features of the fabric. Multidimensional feature extraction is performed using a Gabor filter bank and an improved local binary pattern. The Gabor filter bank can fully capture the pile direction features at different scales through the combination of six scales and eight directions. The improved local binary pattern extracts rotation-invariant texture features using a circular neighborhood sampling strategy. The combination of the two algorithms realizes accurate representation of the texture features of the fabric. The defect analysis method based on the U-Net deep learning model extracts multi-level features by using the ResNet50 backbone network in the encoder, introduces an attention mechanism in the decoder to enhance the perception ability of the defect area, and adds a directional convolution module in the skip connection, which significantly improves the detection sensitivity of the pile direction defects. The genetic algorithm and the double-face pile parameter collaborative control model are used for process parameter optimization. By establishing the mapping relationship between the fabric defect features and the process parameters, adaptive adjustment of the parameters is realized, and the introduction of the genetic algorithm ensures that the optimal solution is found in the complex parameter space. Finally, through comprehensive evaluation of multiple dimensions such as pile density distribution, direction consistency, pile uniformity, fiber softness, surface glossiness and pile height difference, a comprehensive quality evaluation system is established, and the accurate quantification of the fabric processing quality is realized. This intelligent control scheme based on artificial intelligence algorithm not only improves the accuracy and comprehensiveness of the detection, but also realizes closed-loop control of the processing process through real-time feedback and adaptive adjustment mechanism, significantly improving the processing quality and production efficiency of suede super-imitation leather fabric.

[0125] Reference Figure 5 In the embodiment of the present application, a computer device is also provided, which can be a server, and the internal structure thereof can be as shown in Figure 5 The computer device includes a processor, a memory, a display screen, an input device, a network interface and a database connected through a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store the corresponding data in the embodiment. The network interface of the computer device is used to communicate with external terminals through network connection. The computer program is executed by the processor to implement the above method.

[0126] Those skilled in the art can understand,Figure 5 The structure shown in the figure is only a block diagram of part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied.

[0127] The embodiment of the present application further provides a computer readable storage medium, which stores a computer program, and the computer program is executed by a processor to realize the method. It can be understood that the computer readable storage medium in the embodiment can be a volatile readable storage medium or a non-volatile readable storage medium.

[0128] It can be understood by those skilled in the art that all or part of the processes in the above-mentioned embodiment methods can be completed by a computer program instructing related hardware, and the computer program can be stored in a non-volatile computer readable storage medium. When the computer program is executed, the computer program can include the processes of the above-mentioned embodiment methods. Any reference to the memory, storage, database or other medium provided by the present application and used in the embodiments can include non-volatile and / or volatile memory. The non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. The volatile memory can include random access memory (RAM) or external cache memory. As an illustration but not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (SSRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM) and memory bus dynamic RAM, etc.

[0129] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the above-mentioned system, system and unit can refer to the corresponding processes in the above-mentioned method embodiments, which will not be described here.

[0130] The integrated unit, if implemented in the form of a software function unit and sold or used as an independent product, can be stored in a computer readable storage medium. Based on such understanding, the technical solutions of the present application essentially or the part that contributes to the prior art, or all or part of the technical solutions can be embodied in the form of a software product. The computer software product is stored in a storage medium, and includes several instructions to make a computer device (which can be a personal computer, a server, or a network device, etc.) execute all or part of the steps of the methods described in the various embodiments of the present application. The aforementioned storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, and various media that can store program codes.

[0131] The above-described embodiments are merely used to illustrate the technical solutions of the present application, rather than limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that they can still make modifications to the technical solutions recorded in the foregoing embodiments, or make equivalent replacements to some of the technical features; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A processing control method for suede super-imitation leather fabric, characterized by, The processing control method for suede super-imitation leather fabric comprises the following steps: Through a double-light-source alternating illumination system, image acquisition is performed on the suede fabric to obtain complementary image data, and the complementary image data is preprocessed to obtain an enhanced fabric image; Based on a Gabor filter bank and an improved local binary pattern, multi-dimensional feature extraction is performed on the enhanced fabric image to obtain fabric texture feature data; Fabric defect information is obtained by analyzing the fabric texture feature data according to a U-Net deep learning model; Based on a genetic algorithm and a double-sided nap parameter collaborative control model, process parameter calculation is performed on the fabric defect information to obtain process parameter optimization data, including: combining the fabric defect information with nap height, pile density, and fabric tension parameters to construct a parameter association matrix to generate initial process parameters; performing positive and reverse side collaborative analysis on the initial process parameters to establish a matching relationship diagram of double-sided nap rotating speed and carborundum grit number to obtain collaborative parameter data; performing hierarchical mapping on the collaborative parameter data through a fuzzy membership function to obtain a process parameter adjustment sequence; performing iterative updating on the process parameter adjustment sequence based on a reinforcement learning strategy to obtain an optimized parameter combination; performing environment compensation on the optimized parameter combination by steam temperature and fabric humidity to generate a compensated parameter matrix; and performing double-sided balance verification on the compensated parameter matrix to obtain process parameter optimization data; The process parameter optimization data is quality evaluated through pile density distribution, direction consistency, nap uniformity, fiber softness, surface glossiness, and pile height difference to obtain fabric processing quality evaluation results.

2. The processing control method for suede super-imitation leather fabric according to claim 1, characterized in that, The double-light-source alternating illumination system for image acquisition on the suede fabric to obtain complementary image data, and the pre-processing of the complementary image data to obtain an enhanced fabric image comprises the following steps: An original image is generated by acquiring a fabric motion image sequence through a high-speed industrial camera and eliminating motion blur of the image sequence in combination with a motion trajectory compensation algorithm; Through a self-adaptive weight fusion algorithm, surface texture features of orthographic light source data and stereoscopic profile information of oblique light source data are time-series fused to obtain double-light-source complementary image data; Based on image local entropy and gradient distribution, the double-light-source complementary image data is adaptively grid-divided, and the grid size is dynamically determined according to texture complexity to obtain grid-divided data; Through local contrast mapping and global brightness calibration, the grid-divided data is histogram equalized, and the best stretching coefficient is determined through iterative optimization to obtain an equalized image; According to pile direction field distribution, a multi-scale and multi-direction morphological operator is constructed to enhance the pile texture of the equalized image, and edge-preserving filtering is performed to suppress noise to obtain texture enhancement data; Based on a fabric color feature illumination compensation model, the texture enhancement data is nonlinearly corrected and contrast optimized, and a parameter self-adaptive adjustment is performed in combination with an illumination unevenness evaluation index to obtain an enhanced fabric image.

3. The processing control method for suede super-imitation leather fabric according to claim 1, characterized in that, The multi-dimensional feature extraction on the enhanced fabric image based on a Gabor filter bank and an improved local binary pattern to obtain fabric texture feature data comprises the following steps: The enhanced fabric image is subjected to direction feature analysis by constructing a six-scale eight-direction Gabor filter bank to obtain pile direction feature data; The pile direction feature data is subjected to four-level decomposition based on a sym4 wavelet base to extract multi-scale texture coefficients of 600-mesh and 400-mesh double-sided pile; The multi-scale texture coefficients are subjected to binary mapping by circular neighborhood sampling to generate a rotation-invariant feature sequence of the front and back surfaces; Energy, entropy and contrast feature values are calculated for the rotation-invariant feature sequence to construct a double-sided texture gray statistical matrix; The gray statistical matrix is subjected to humidity compensation correction according to a 25-30% humidity parameter to obtain a compensated feature vector; The compensated feature vector is subjected to weighted fusion with the double-sided pile direction feature data to generate fabric texture feature data.

4. The processing control method for suede super-imitation leather fabric according to claim 1, characterized in that, The fabric texture feature data is analyzed according to a UNet deep learning model to obtain fabric defect information, including: The fabric texture feature data is subjected to hierarchical feature extraction by a multi-layer convolution progressive network to obtain a hierarchical feature mapping; The hierarchical feature mapping is subjected to double-weighted operation based on a channel attention mechanism and a spatial attention mechanism to obtain an enhanced feature map; The enhanced feature map is subjected to pile texture analysis by a multi-direction convolution kernel group to generate front and back surface direction defect features; The front and back surface direction defect features are subjected to pixel-level depth segmentation and boundary constraint detection to obtain double-sided defect data; The double-sided defect data is subjected to environmental factor correction by a pressure coefficient and a steam parameter to obtain a compensated defect matrix; The compensated defect matrix is subjected to defect classification in combination with defect morphological features and texture statistical features to generate fabric defect information.

5. The method for processing control of suede super-imitation leather fabric according to claim 1, characterized in that, The process parameter optimization data is subjected to quality evaluation by pile density distribution, direction consistency, pile uniformity, fiber softness, surface glossiness and pile height difference to obtain fabric processing quality evaluation results, including: The fabric sample is subjected to region segmentation by the process parameter optimization data to obtain evaluation unit data, and the evaluation unit data is subjected to statistical operation by unit area pile count to generate pile density distribution; The evaluation unit data is subjected to front and back surface scanning, and direction consistency is calculated according to pile orientation angle deviation, and pile uniformity is calculated by region density variance analysis on the evaluation unit data; The evaluation unit data is subjected to pressure response measurement, and fiber softness is calculated according to the pressure deformation curve; The evaluation unit data is subjected to multi-angle optical scanning to measure reflection intensity and generate surface glossiness; The evaluation unit data is subjected to double-sided height measurement, and pile height difference is calculated according to the front and back surface height difference; The pile density distribution, direction consistency, pile uniformity, fiber softness, surface glossiness and pile height difference are subjected to weighted calculation to obtain fabric processing quality evaluation results.

6. The method for processing control of suede super-imitation leather fabric according to claim 1, characterized in that, The processing control method for suede super-imitation leather fabric further includes: The fabric processing quality evaluation results are subjected to data hierarchical scanning by a quality correction probe to generate quality distribution surface data; The mass distribution surface data is subjected to time sequence comparison operation, and a dynamic control sequence is calculated according to the mass fluctuation curve; The dynamic control sequence is subjected to working condition analysis through a multi-point sensing network, and a process deviation matrix is obtained; The process deviation matrix is subjected to partition operation, a deviation compensation curve is generated, process parameters are dynamically configured according to the deviation compensation curve, and a parameter adjustment sequence is obtained; A process control link is constructed according to the parameter adjustment sequence, control instructions are decomposed for the fabric processing equipment, and equipment control data flow is generated; The equipment control data flow is used for cooperative scheduling of the process execution unit, execution feedback data is obtained, and quality optimization control instructions are generated through closed-loop iterative calculation of the execution feedback data and the fabric processing quality evaluation result.

7. A processing control system for suede super-imitation leather fabric for implementing the processing control method for suede super-imitation leather fabric according to any one of claims 1 to 6, characterized by, The processing control system for suede super imitation leather fabric includes: The acquisition module is used for image acquisition of suede fabric through a double-light-source alternating illumination system to obtain complementary image data, and the complementary image data is preprocessed to obtain enhanced fabric images. The extraction module is used for multi-dimensional feature extraction of the enhanced fabric images based on a Gabor filter bank and an improved local binary pattern to obtain fabric texture feature data. The analysis module is used for analysis of the fabric texture feature data according to a U-Net deep learning model to obtain fabric defect information. The calculation module is used for process parameter calculation of the fabric defect information based on a genetic algorithm and a double-sided nap parameter cooperative control model to obtain process parameter optimization data, including: combining the fabric defect information with nap height, pile density, and fabric tension parameters to construct a parameter association matrix to generate initial process parameters; performing positive and reverse side cooperative analysis on the initial process parameters to establish a matching relationship diagram of double-sided nap rotating speed and diamond grit number to obtain cooperative parameter data; performing hierarchical mapping on the cooperative parameter data through a fuzzy membership function to obtain a process parameter adjustment sequence; performing iterative updating on the process parameter adjustment sequence based on a reinforcement learning strategy to obtain an optimized parameter combination; performing environmental compensation on the optimized parameter combination with steam temperature and cloth surface humidity to generate a compensated parameter matrix; performing double-sided balance verification on the compensated parameter matrix to obtain process parameter optimization data. The evaluation module is used for quality evaluation of the process parameter optimization data through pile density distribution, direction consistency, nap uniformity, fiber softness, surface glossiness, and nap height difference to obtain fabric processing quality evaluation results.

8. A computer device, comprising: A memory and a processor are included, the memory stores a computer program capable of running on the processor, and the processor implements the processing control method for suede super imitation leather fabric according to any one of claims 1-6 when executing the computer program. 9.A computer readable storage medium having a computer program stored thereon, the computer program, when executed by a processor, causing the processor to perform the processing control method for suede super imitation leather fabric according to any one of claims 1-6.

Images