Intelligent control methods and systems for automated textile equipment

Abstract

This application discloses an intelligent control method for automated textile equipment, comprising the following steps: Step 1: Obtain the textile pattern to be woven, extract feature points from the textile pattern, and obtain a feature point set; Step 2: Perform discrete cosine transform on the textile pattern to transform the textile pattern from the spatial domain to the frequency domain, and obtain the frequency domain coefficient matrix of the textile pattern; Step 3: Generate a flag matrix corresponding to the frequency domain coefficient matrix, wherein each element in the flag matrix and the frequency domain coefficient matrix corresponds one-to-one; In the technical solution provided in this application, the actual textile pattern is monitored by real-time image sensing technology, and the feature points are compared with the original design pattern to accurately calculate the dynamic offset between the actual position and the expected position of the needle (or knitting needle).

Description

Technical Field

[0001] This application relates to the field of image processing technology, and more specifically, to an intelligent control method and system for automated textile equipment. Background Technology

[0002] Smart textile technology can automatically generate textile pattern codes through programming and drive textile equipment to control the movement of the fabric, using different colored threads to weave the desired patterns on the fabric.

[0003] This process typically relies on a transport mechanism to move the fabric under the needles. However, in practice, this weaving method has a key problem: although the transport mechanism is driven by an electric motor with precise speed control, the fabric movement mainly depends on friction. This results in uneven fabric speed during movement. Consequently, the actual placement of the needles deviates from the intended position, ultimately reducing the accuracy of the woven pattern. Summary of the Invention

[0004] The summary section of this application is intended to provide a brief overview of the concepts, which will be described in detail in the detailed description section below. This summary section is not intended to identify key or essential features of the claimed technical solutions, nor is it intended to limit the scope of the claimed technical solutions.

[0005] As a first aspect of this application, in order to solve the technical problems mentioned in the background section above, some embodiments of this application provide an intelligent control method for automated textile equipment, including the following steps:

[0006] Step 1: Obtain the textile pattern to be woven, extract feature points from the textile pattern, and obtain a set of feature points;

[0007] Step 2: Perform discrete cosine transform on the textile pattern to transform it from the spatial domain to the frequency domain and obtain the frequency domain coefficient matrix of the textile pattern.

[0008] Step 3: Generate a flag matrix that corresponds to the frequency domain coefficient matrix, with each element in the flag matrix and the frequency domain coefficient matrix having a one-to-one correspondence;

[0009] Step 4: Mark the frequency domain coefficient matrix with a label matrix to obtain the labeled frequency domain matrix;

[0010] Step 5: Real-time detection of the monitoring patterns collected during the textile process, random selection of several feature points from the monitoring patterns, acquisition of the corresponding positions of the feature points in the marker matrix, and determination of the offset between the current needle position and the actual position based on the offset of the corresponding position.

[0011] Step 6: Adjust the weaving scheme according to the offset between the current needle position and the actual position.

[0012] Furthermore, S01: Extract feature points from textile patterns based on the SIFT algorithm; specifically:

[0013] For a textile pattern I(x,y), the scale space L(x,y,σ) is defined as: L(x,y,σ) = G(x,y,σ) × I(x,y);

[0014] Where G(x, y, σ) is a Gaussian function;

[0015] ;

[0016] S02: For each feature point, one or more principal directions are determined based on the gradient orientation histogram in the neighborhood of the key point.

[0017] The gradient magnitude m and direction θ of the feature point are as follows:

[0018] ;

[0019] ;

[0020] Where x and y are the x and y coordinates of the pixel, respectively, σ is the standard deviation of the Gaussian function, m(x,y) represents the gradient magnitude at position (x,y), and L refers to the image after Gaussian blurring of image I at different scales.

[0021] Furthermore, in step 1, feature points are extracted based on a neural network model. The neural network model learns and integrates the correspondence between features and labels, and automatically labels each region.

[0022] Furthermore, step 1 includes the following steps:

[0023] Step 11: Divide the textile pattern into multiple rectangular areas.

[0024] Step 12: The iterative unit randomly selects a region i and selects a segmentation label based on the state information of region i. The state information includes the pixel information of the region and the pixel information of adjacent regions.

[0025] Step 13: Calculate the reward value r for the segmentation label assigned to region i.

[0026] Furthermore, the status information includes S wi S ei S li ;

[0027] For region i, its pixel information is S wi The pixel information of the remaining regions adjacent to region i is S. eiThe label distribution of the remaining adjacent regions of region i is S. li ;

[0028] S ei ={E 1i E 2i E 3i …E ei …}, where E ei It is the pixel information of the e-th region among the remaining regions adjacent to region i;

[0029] S li ={L 1i L 2i L 3i …L li …}, L li It is a distribution vector, where the l-th element represents the label of the l-th region among the remaining regions adjacent to region i.

[0030] In the technical solution provided in this application, the label of the region is updated based on the pixel information of the region and the labels of adjacent regions. Therefore, relatively speaking, less annotation data is required, which increases the accuracy of the region labeling and allows for better identification of the relationship between information during the iteration process.

[0031] When training a model, labeled data is required. However, this labeled data is not always accurate; it contains many erroneous data points that are difficult to distinguish. Using this erroneous data in the training of the labeling module can lead to the module learning incorrect information. Therefore, this application provides the following technical solution:

[0032] Furthermore, after all regions are labeled, the loop unit filters out regions whose labels differ from those of the surrounding regions and inputs them into the iteration unit for re-iteration.

[0033] The technical solution provided in this application selects regions that may have mislabeled labels based on the distribution of labels between regions, and then guides these regions to undergo further reinforcement learning, thereby reducing the impact of erroneous data on prediction accuracy.

[0034] Furthermore, during the iteration process, the model is affected by the reward value, and the design of the reward value affects the convergence rate of the model and increases the training cost. To address this, this application provides the following technical solution:

[0035] Furthermore, the reward value is r. ;

[0036] Let be a Gaussian function, min(d)i ) represents the region i to the true value partition boundary G. i The shortest geodesic distance among all points, where the geodesic distance is the length of the shortest path from one point to another, G. i This refers to the correct segmentation result for region i, and the distance from region i to the true value segmentation boundary G. i The minimum geodesic distance is the label of region i, and α and β are a pair of scale parameters.

[0037] Furthermore, the training process of the neural network model is as follows:

[0038] S1: Initialize the parameters of the RLSegNet neural network model. This includes setting the number of iterations, step size, α, and β.

[0039] S2: For each training image, segment it into regions and initialize the label of each region to empty or random label.

[0040] S3: Start iteration:

[0041] For each region i, calculate its fusion feature S. wi .

[0042] The average fusion feature S of neighboring regions of region i is calculated. ei .

[0043] Calculate the label distribution S of adjacent regions of region i li .

[0044] S wi、 S ei S li The state information is input into the RLSegNet network, and the network outputs the predicted label for region i.

[0045] S4: For each region i, the segmentation boundary G is determined based on its predicted label and ground truth value. i Calculate the minimum geodesic distance min(d) i ).

[0046] The reward value r is calculated using a Gaussian function, where α=1, β=1, and the cumulative reward value R is calculated.

[0047] S5: Update the parameters of the RLSegNet network using the backpropagation algorithm based on the cumulative reward value R and the results of the reward calculation unit.

[0048] Furthermore, the frequency domain coefficient matrix of the textile pattern is obtained by performing the following algorithm:

[0049] ;

[0050] Where u and v are the horizontal and vertical coordinates in the frequency coefficient matrix, respectively, C(u, v) represents the frequency domain information at (x, y) in the image, M and N are the number of rows and columns of pixels in the textile pattern, respectively, and a(u) and a(v) are normalization factors.

[0051] Furthermore, step 5 includes the following steps:

[0052] Step 51: Use predefined mapping functions ;

[0053] ;

[0054] W*H: Original textile pattern resolution, M*N: DCT frequency domain matrix dimension Represents the spatial coordinate components of the feature point.

[0055] Step 52: Perform DCT transformation on the real-time acquired monitoring pattern:

[0056] , Represents the DCT coefficient matrix of the monitoring pattern;

[0057] Obtain the coefficients at a specific frequency domain location:

[0058] ;

[0059] Represents frequency domain coordinates, Represents a specific frequency domain coefficient value;

[0060] Step 53: Perform XOR positioning using the pre-stored flag matrix:

[0061] ;

[0062] K represents the flag matrix, This represents the XOR operator. Represents the frequency domain coefficients after conversion;

[0063] Step 54: Calculate the expected location using local inverse DCT;

[0064] ;

[0065] ;

[0066] This represents the expected position value of the k-th feature point. Represents the DCT normalization coefficient;

[0067] Step 55: Calculate the offset based on the current expected position and the actual position;

[0068] ;

[0069] This represents the 3D offset of the k-th feature point. Indicates the X-axis position offset, Indicates the Y-axis position offset, Indicates the rotation angle offset;

[0070] Step 6 includes the following steps:

[0071] Step 61: Establish a control parameter correction model;

[0072] ;

[0073] The linear speed of the fabric conveying mechanism after correction;

[0074] The needle mechanism corrects the angular velocity;

[0075] : Proportional / derivative gain matrix of PID controller;

[0076] Step 62: [The text appears to be incomplete and contains several grammatical errors. A more accurate translation would require the full context.] , The commands are converted into motor control instructions and sent to the actuator via real-time Ethernet, with parameter updates completed before the start of the next pin cycle.

[0077] The beneficial effects of this application are as follows:

[0078] Significantly improves the accuracy of textile patterns: By monitoring the actual textile pattern through real-time image sensing technology and comparing its feature points with the original design pattern, the dynamic offset between the actual and expected positions of the needles (or knitting needles) is accurately calculated. Based on this offset, the textile scheme (such as fabric feed speed or needle trajectory) is adjusted in real time, effectively compensating for positioning errors caused by uneven fabric movement (such as friction slippage), thereby greatly improving the geometric accuracy and detail reproduction of the final textile pattern.

[0079] Enhancing System Anti-interference Capability and Stability: This method does not rely on the ideal uniform motion assumption of the transport mechanism itself. By actively sensing and compensating for unavoidable speed fluctuations and positional drifts during fabric movement, it significantly improves the stability and reliability of intelligent textile equipment under actual working conditions (where there are disturbances such as friction and tension changes).

[0080] Achieving closed-loop intelligent control: Image sensing, feature matching, offset calculation, and actuator adjustment are tightly integrated to form a real-time closed-loop feedback control system. The system can "sense, judge, and adjust," enabling the textile process to be adaptive and reducing excessive reliance on the precision of mechanical transmission.

[0081] Optimizing production quality and efficiency: High-precision patterned weaving reduces the rate of defective and scrap products caused by misalignment. Simultaneously, real-time adjustment avoids downtime for corrections due to excessive accumulated errors, contributing to improved production efficiency and yield, and lower production costs.

[0082] Provides a reliable offset detection mechanism: The original pattern information is processed using Discrete Cosine Transform (DCT) and a unique marker matrix encryption method, which provides a structured and more interference-resistant data foundation for subsequent real-time monitoring of pattern feature point extraction and matching. This helps to accurately and efficiently identify feature points and calculate offsets against complex textile texture backgrounds. Attached Figure Description

[0083] The accompanying drawings, which form part of this application, are used to provide a further understanding of the application and to make other features, objects, and advantages of the application more apparent. The illustrative embodiments and descriptions of this application are used to explain the application and do not constitute an undue limitation of the application.

[0084] Furthermore, throughout the accompanying drawings, the same or similar reference numerals denote the same or similar elements. It should be understood that the drawings are schematic, and the elements are not necessarily drawn to scale.

[0085] In the attached diagram:

[0086] Figure 1 This is a flowchart of an intelligent control method for automated textile equipment. Detailed Implementation

[0087] Embodiments of this application will now be described in more detail with reference to the accompanying drawings. While some embodiments of this application are shown in the drawings, it should be understood that this application can be implemented in various forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided to provide a more thorough and complete understanding of this application. It should be understood that the drawings and embodiments of this application are for illustrative purposes only and are not intended to limit the scope of protection of this application.

[0088] It should also be noted that, for ease of description, only the parts relevant to the invention are shown in the accompanying drawings. Unless otherwise specified, the embodiments and features described herein can be combined with each other.

[0089] The present application will now be described in detail with reference to the accompanying drawings and embodiments.

[0090] The intelligent control method for automated textile equipment includes the following steps:

[0091] Step 1: Obtain the textile pattern to be woven, extract feature points from the textile pattern, and obtain a set of feature points.

[0092] Step 1 primarily aims to extract feature points from the textile pattern. These feature points need to be hidden, undiscoverable, and resistant to transformations. In other words, they should not change even after image sharpness alterations or rotations. Generally, the SIFT algorithm is used to extract image feature points.

[0093] The specific solution is as follows: This application uses the following solution to extract feature points:

[0094] S01: Extracting feature points from textile patterns based on the SIFT algorithm; specifically:

[0095] For a textile pattern I(x,y), the scale space L(x,y,σ) is defined as: L(x,y,σ) = G(x,y,σ) × I(x,y);

[0096] Where G(x, y, σ) is a Gaussian function;

[0097] ;

[0098] S02: For each feature point, one or more principal directions are determined based on the gradient orientation histogram in the neighborhood of the key point.

[0099] The gradient magnitude m and direction θ of the feature point are as follows:

[0100] ;

[0101] ;

[0102] Where x and y are the x and y coordinates of the pixel, respectively, σ is the standard deviation of the Gaussian function, m(x,y) represents the gradient magnitude at position (x,y), and L refers to the image after Gaussian blurring of image I at different scales.

[0103] The above scheme can obtain all feature points in the textile pattern, thus yielding a feature point set. However, this scheme requires applying the SIFT algorithm to the entire image when extracting feature points, resulting in low encryption efficiency. Therefore, this application provides a neural network model for feature point extraction.

[0104] Specifically, step 1 includes the following steps:

[0105] Step 11: Divide the textile pattern into multiple rectangular areas.

[0106] Step 12: The iterative unit randomly selects a region i and selects a segmentation label based on the state information of region i. The state information includes the pixel information of the region and the pixel information of adjacent regions.

[0107] Step 13: Calculate the reward value r for the segmentation label assigned to region i;

[0108] The reward value is essentially an adaptive function. The larger the reward value r, the higher the reward of the iteration unit, and vice versa.

[0109] The loop unit guides the iterative unit to continuously loop until the maximum cumulative reward value R is obtained.

[0110] In this scheme, the kernel of the iterative unit is the hidden layer of the neural network model to be trained, which is the RLSegNet network. The recurrent unit controls the number of iterations during training, and the reward calculation unit is used to determine the validity of the current iteration operation.

[0111] Specifically, the labels in this scheme are actually the positions of feature points within a region. For example, if the region size in this scheme is 10*10, the labels are an identifier matrix indicating whether a pixel is a feature point. Here, 0 indicates that it is not a feature point, and 1 indicates that it is a feature point.

[0112] During training, a large number of samples are needed. Whether a sample contains a feature point is determined by the aforementioned feature point extraction algorithm.

[0113] Furthermore, within the iterative unit:

[0114] Status information includes S wi S ei S li ;

[0115] For region i, its pixel information is S wi The pixel information of the remaining regions adjacent to region i is S. ei The label distribution of the remaining adjacent regions of region i is S. li ;

[0116] S ei ={E 1i E 2i E 3i …E ei …}, where E ei It is the pixel information of the e-th region among the remaining regions adjacent to region i;

[0117] S li ={L1i L 2i L 3i …L li …}, L li It is a distribution vector, where the l-th element represents the label of the l-th region among the remaining regions adjacent to region i;

[0118] Adjacent areas generally refer to areas that are adjacent to other areas. In this plan, it mainly refers to areas within the first ring road, that is, areas directly adjacent to other areas. In practice, it can also refer to areas within the second ring road, that is, indirectly adjacent areas.

[0119] Furthermore, after all regions are labeled, the loop unit filters out regions whose labels differ from those of the surrounding regions and inputs them into the iteration unit for re-iteration.

[0120] Furthermore, during the iteration process, the model is affected by the reward value, and the design of the reward value affects the convergence rate of the model and increases the training cost. To address this, this application provides the following technical solution:

[0121] Furthermore, the reward value is r. ;

[0122] Let be a Gaussian function, and min(di) represent the distance from region i to the truth boundary G. i The shortest geodesic distance among all points, where the geodesic distance is the length of the shortest path from one point to another, G. i This refers to the correct segmentation result for region i. The minimum geodesic distance from region i to the true segmentation boundary Gi is the label of region i. α and β are a pair of scaling parameters, and r... t This represents the index at the t-th iteration, where t represents the index of the iteration number, and a t This represents the action at the t-th iteration, so a t =G i This indicates that the segmentation result is correct; otherwise, it is incorrect. In this scheme, α=1 and β=1.

[0123] Specifically, the training process is as follows:

[0124] S1: Initialize the parameters of the RLSegNet neural network model. This includes setting the number of iterations, step size, α, and β.

[0125] S2: For each training image, segment it into regions and initialize the label of each region to empty or random label.

[0126] S3: Start iteration:

[0127] For each region i, calculate its fusion feature S.wi .

[0128] The average fusion feature S of neighboring regions of region i is calculated. ei .

[0129] Calculate the label distribution S of adjacent regions of region i li .

[0130] S wi、 S ei S li The state information is input into the RLSegNet network, and the network outputs the predicted label for region i.

[0131] S4: For each region i, the segmentation boundary G is determined based on its predicted label and ground truth value. i Calculate the minimum geodesic distance min(d) i ).

[0132] The reward value r is calculated based on the Gaussian function, where α=1, β=1, and the cumulative reward value R.

[0133] Loop unit operations:

[0134] Check if all regions have been labeled, filter out regions with labels different from their surrounding regions, and re-enter these regions into the iteration unit for re-iteration. Repeat the iteration process until the maximum number of iterations is reached or the cumulative reward value R no longer increases significantly.

[0135] S5: Based on the cumulative reward value R and the results of the reward calculation unit, update the parameters of the RLSegNet network using the backpropagation algorithm. Continue until the RLSegNet network converges or reaches the preset number of training epochs: Save the trained RLSegNet model parameters.

[0136] To increase the convergence speed of the model:

[0137] In this scheme, α and β are dynamically changing and are related to the number of iterations. Let the current number of iterations be t, and the total number of iterations be T.

[0138] ;

[0139] Here, U and k are pre-set control parameters, with U=3 and k=1. Thus, in this scheme, at the beginning of the iteration, the growth rate of the reward function is low, meaning both the reward and penalty are small. Therefore, the iterative unit has more choices during iteration, preventing the model from getting trapped in local optima. However, in the latter half of the iteration, the growth rate of the reward function will increase. At this point, incorrect choices will result in larger rewards or penalties. Therefore, when selecting iteration parameters, we will try to choose more correct operations to increase the model's convergence rate.

[0140] When making predictions:

[0141] (1): Load the parameters of the trained RLSegNet model.

[0142] (2): Segment the image to be predicted into regions and calculate the fusion feature S of each region. wi .

[0143] (3): For each region i, initialize its label to empty or random label, and calculate the average fusion feature S of the neighboring regions of region i. ei and label distribution S li , will S wi S ei S li The state information is input into the RLSegNet network, which outputs the predicted label for region i. This process is repeated iteratively until all regions are labeled or a preset number of iterations is reached. Post-processing of the prediction results, such as smoothing boundaries and removing outliers, is performed to improve the accuracy of the segmentation results.

[0144] Output segmentation results: Integrate the prediction results into a complete image segmentation map and output it.

[0145] Step 2: Perform discrete cosine transform on the textile pattern to transform it from the spatial domain to the frequency domain, and obtain the frequency domain coefficient matrix of the textile pattern.

[0146] Specifically:

[0147] ;

[0148] Where u and v are the horizontal and vertical coordinates in the frequency coefficient matrix, respectively, C(u, v) represents the frequency domain information at (x, y) in the image, M and N are the number of rows and columns of pixels in the textile pattern, respectively, and a(u) and a(v) are normalization factors.

[0149] Step 2 transforms the textile pattern from the spatial domain to the frequency domain, thus highlighting the frequency domain information more prominently.

[0150] Step 3: Generate a flag matrix that corresponds to the frequency domain coefficient matrix. Each element in the flag matrix and the frequency domain coefficient matrix have a one-to-one correspondence.

[0151] Step 4: Mark the frequency domain coefficient matrix with a label matrix to obtain the labeled frequency domain matrix;

[0152] After obtaining the frequency domain coefficient matrix of the textile pattern, all information about the textile pattern can be represented using the frequency domain coefficient matrix. In the frequency domain, a transformation operation is performed on the transform coefficients.

[0153] C′ (u,v)=C(u,v)⊕K(u,v)

[0154] Where C(u,v) are the transformed frequency domain coefficients, K(u,v) is the flag matrix, ⊕ represents the XOR operation, and C′(u,v) are the encrypted frequency domain coefficients.

[0155] Step 5: Real-time detection of the monitoring patterns collected during the textile process, random selection of several feature points from the monitoring patterns, acquisition of the corresponding positions of the feature points in the marker matrix, and determination of the offset between the current needle position and the actual position based on the offset of the corresponding position.

[0156] Specifically, images of the fabric surface under the needles are captured in real time using a high frame rate industrial camera and recorded as monitoring patterns. The image acquisition frequency is ≥ twice the needle movement frequency of the textile equipment (satisfying Shannon's sampling theorem).

[0157] In step 5, the collected monitoring images are compared with the textile images using the following method for detection;

[0158] Step 5 includes the following steps:

[0159] Step 51: Use predefined mapping functions ;

[0160] ;

[0161] W*H: Original textile pattern resolution, M*N: DCT frequency domain matrix dimension Represents the spatial coordinate components of the feature point.

[0162] Step 52: Perform DCT transformation on the real-time acquired monitoring pattern:

[0163] , Represents the DCT coefficient matrix of the monitoring pattern;

[0164] Obtain the coefficients at a specific frequency domain location:

[0165] ;

[0166] Represents frequency domain coordinates, Represents a specific frequency domain coefficient value;

[0167] Step 53: Perform XOR positioning using the pre-stored flag matrix:

[0168] ;

[0169] K represents the flag matrix, This represents the XOR operator. Represents the frequency domain coefficients after conversion;

[0170] Step 54: Calculate the expected location using local inverse DCT;

[0171] ;

[0172] ;

[0173] This represents the expected position value of the k-th feature point. Represents the DCT normalization coefficient;

[0174] Step 55: Calculate the offset based on the current expected position and the actual position;

[0175] ;

[0176] This represents the 3D offset of the k-th feature point. Indicates the X-axis position offset, Indicates the Y-axis position offset, This indicates the rotation angle offset.

[0177] Step 6: Adjust the weaving scheme according to the offset between the current needle position and the actual position.

[0178] Step 6 includes the following steps:

[0179] Step 61: Establish a control parameter correction model;

[0180] ;

[0181] The linear speed of the fabric conveying mechanism after correction;

[0182] The needle mechanism corrects the angular velocity;

[0183] : Proportional / derivative gain matrix of PID controller;

[0184] Step 62: [The text appears to be incomplete and contains several grammatical errors. A more accurate translation would require the full context.] , The commands are converted into motor control instructions and sent to the actuator via real-time Ethernet, with parameter updates completed before the start of the next pin cycle.

[0185] The above description is merely a selection of preferred embodiments of this application and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of the invention involved in the embodiments of this application is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the above-described inventive concept. For example, technical solutions formed by substituting the above-described features with (but not limited to) technical features with similar functions disclosed in the embodiments of this application.

Claims

1. An intelligent control method for automated textile equipment, characterized in that: Includes the following steps: Step 1: Obtain the textile pattern to be woven, extract feature points from the textile pattern, and obtain a set of feature points; Step 2: Perform discrete cosine transform on the textile pattern to transform it from the spatial domain to the frequency domain and obtain the frequency domain coefficient matrix of the textile pattern. Step 3: Generate a flag matrix that corresponds to the frequency domain coefficient matrix, with each element in the flag matrix and the frequency domain coefficient matrix having a one-to-one correspondence; Step 4: Mark the frequency domain coefficient matrix with a label matrix to obtain the labeled frequency domain matrix; Step 5: Real-time detection of the monitoring patterns collected during the textile process, random selection of several feature points from the monitoring patterns, acquisition of the corresponding positions of the feature points in the marker matrix, and determination of the offset between the current needle position and the actual position based on the offset of the corresponding position. Step 6: Adjust the weaving plan according to the offset between the current needle position and the actual position; Step 5 includes the following steps: Step 51: Use predefined mapping functions ; ； W*H: Original textile pattern resolution; M*N: Dimensions of the DCT frequency domain coefficient matrix. Represents the spatial coordinate components of the feature point; Step 52: Perform DCT transformation on the real-time acquired monitoring pattern: , Represents the DCT coefficient matrix of the monitoring pattern; Obtain the coefficients at a specific frequency domain location: ； Represents frequency domain coordinates, Represents a specific frequency domain coefficient value; Step 53: Perform XOR positioning using the pre-stored flag matrix: ； K represents the flag matrix, This represents the XOR operator. Represents the frequency domain coefficients after conversion; Step 54: Calculate the expected location using local inverse DCT; ； ； This represents the expected position value of the k-th feature point. Represents the DCT normalization coefficient; Step 55: Calculate the offset based on the current expected position and the actual position; ； This represents the 3D offset of the k-th feature point. Indicates the X-axis position offset, Indicates the Y-axis position offset, This indicates the rotation angle offset.

2. The intelligent control method for automated textile equipment according to claim 1, characterized in that: S01: Extracting feature points from textile patterns based on the SIFT algorithm; specifically: For a textile pattern I(x,y), the scale space L(x,y,σ) is defined as: L(x,y,σ) = G(x,y,σ) × I(x,y); Where G(x, y, σ) is a Gaussian function; S02: For each feature point, one or more principal directions are determined based on the gradient orientation histogram in the neighborhood of the key point. The gradient magnitude m and direction θ of the feature point are as follows: ； ； Where x and y are the x and y coordinates of the pixel, respectively, σ is the standard deviation of the Gaussian function, m(x,y) represents the gradient magnitude at position (x,y), and L refers to the image after Gaussian blurring of image I at different scales.

3. The intelligent control method for automated textile equipment according to claim 2, characterized in that: In step 1, feature points are extracted based on a neural network model. The neural network model learns and integrates the correspondence between features and labels, and automatically labels each region.

4. The intelligent control method for automated textile equipment according to claim 3, characterized in that: Step 1 includes the following steps: Step 11: Divide the textile pattern into multiple rectangular areas; Step 12: The iterative unit randomly selects a region i, and selects a segmentation label based on the state information of region i. The state information includes the pixel information of the region and the pixel information of adjacent regions; Step 13: Calculate the reward value r for the segmentation label assigned to region i.

5. The intelligent control method for automated textile equipment according to claim 4, characterized in that: Status information includes S wi S ei S li ; For region i, its pixel information is S wi The pixel information of the remaining regions adjacent to region i is S. ei The label distribution of the remaining adjacent regions of region i is S. li ; S ei ={E 1i E 2i E 3i …E ei …}, where E ei It is the pixel information of the e-th region among the remaining regions adjacent to region i; S li ={L 1i L 2i L 3i …L li …}, L li It is a distribution vector, where the l-th element represents the label of the l-th region among the remaining regions adjacent to region i.

6. The intelligent control method for automated textile equipment according to claim 4, characterized in that: After all regions are labeled, the loop unit filters out regions whose labels differ from those of the surrounding regions and inputs them into the iteration unit for re-iteration.

7. The intelligent control method for automated textile equipment according to claim 6, characterized in that: The reward value is r. ; Let be a Gaussian function, min(d) i ) represents the region i to the true value partition boundary G. i The shortest geodesic distance among all points, where the geodesic distance is the length of the shortest path from one point to another, G. i This refers to the correct segmentation result of region i, where region i is the distance from the ground truth segmentation boundary G. i The minimum geodesic distance is the label of region i, and α and β are a pair of scale parameters.

8. The intelligent control method for automated textile equipment according to claim 7, characterized in that: The following algorithm is used to obtain the frequency domain coefficient matrix of the textile pattern: ； Where u and v are the horizontal and vertical coordinates in the frequency coefficient matrix, respectively, C(u, v) represents the frequency domain information at (x, y) in the image, M and N are the number of rows and columns of pixels in the textile pattern, respectively, and a(u) and a(v) are normalization factors.

9. The intelligent control method for automated textile equipment according to claim 7, characterized in that: Step 6 includes the following steps: Step 61: Establish a control parameter correction model; ； The linear speed of the fabric conveyor mechanism after correction; The needle mechanism corrects the angular velocity; : Proportional / derivative gain matrix of PID controller; Step 62: [The text appears to be incomplete and contains several grammatical errors. A more accurate translation would require the , The commands are converted into motor control instructions and sent to the actuator via real-time Ethernet, with parameter updates completed before the start of the next pin cycle.

Images