Cloth paving machine visual edge alignment control method for complex textures and composite fabrics
By using vision system and grayscale symbiosis matrix technology in the laying machine, the opposite edge positions of composite fabrics and complex texture fabrics are accurately obtained, and the problems of low edge accuracy and large oscillation are solved, and high-precision opposite edge control is achieved.
Patent Information
- Application Number
- CN202510095621.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-01-21
- Publication Date
- 2025-05-16
AI Technical Summary
When existing laying machines deal with composite fabrics and fabrics with complex textures, it is difficult to accurately obtain the edge position of the edge, resulting in low accuracy and large oscillation.
The visual system is used to combine the color processing methods of the grayscale symbiosis matrix to divide the edge areas, calculate the standard feature quantity and the actual feature quantity, judge the edge area type through the feature measurement value, and output the reference boundary pixel position, and calculate the position error value to control the action of the edge mechanism.
Accurate side-to-side control of composite fabrics and complex texture fabrics is achieved, edge-to-side precision is improved, oscillation is reduced, and the accuracy exceeds the traditional photoelectric detection.
Smart Images
Figure CN120013903A_ABST
Abstract
Description
Technical Field
[0001] The invention relates to the technical field of cloth spreading machines, and in particular to a visual edge control method for a cloth spreading machine. Background Art
[0002] In the application of the spreading machine, the conventional spreading process needs to meet the requirements of longitudinal and transverse edge alignment at the same time, among which the edge alignment is mainly completed by the servo motor driving the spreading machine to move through the transmission device.
[0003] In the edge alignment servo drive, the position feedback signal is generally monitored by a photoelectric sensor. The photoelectric sensor can detect whether the emitted infrared beam is blocked by the cloth through reflected light, and output a switch signal to control and drive the servo motor to move in both the horizontal and vertical directions. Although the photoelectric sensor can determine the position range of the edge of the cloth, it cannot give an accurate edge position feedback signal, resulting in large oscillations in the edge alignment, affecting the edge alignment accuracy. To address this problem, in the prior art, some cloth spreading machines use a visual system to obtain the fabric edge alignment lines, and drive the edge alignment mechanism through a servo system, which can effectively improve the edge alignment accuracy and reduce oscillations during the edge alignment process.
[0004] However, as the fabrics being laid become increasingly complex, on the one hand, some composite fabrics and fabrics with raw edges generally use the inner edge as the actual sewing object when laying the fabrics; on the other hand, the complex color and texture of the fabrics also make it difficult to distinguish the edge targets, making it difficult for traditional vision-based edge detection systems to effectively deal with this situation. Summary of the invention
[0005] The purpose of the present invention is to provide a visual edge control method for a spreading machine for complex textured composite fabrics. The method can accurately obtain the edge position of the spreading machine and precisely control the edge action of the spreading machine to effectively solve the edge problem of complex textures and composite fabrics.
[0006] In order to achieve the above object, the technical solution adopted by the present invention is as follows:
[0007] A visual edge control method for a spreading machine for complex textures and composite fabrics, characterized by comprising the following steps:
[0008] S1, dividing the opposite edge regions, obtaining a sampling pixel region in each opposite edge region, and calculating the standard feature quantity of each sampling pixel region;
[0009] S2, start spreading, select a pixel area to be detected and calculate its actual feature quantity;
[0010] S3, calculating a feature measurement value according to the standard feature quantity and the actual feature quantity; judging the type of the opposite side region into which the currently selected pixel region to be detected falls according to the feature measurement value;
[0011] S4, if it falls into the non-target opposite edge area, automatically select the next pixel area to be detected along the arrangement direction of the opposite edge area and judge again until the selected pixel area to be detected falls into the target opposite edge area, and output the reference boundary pixel position of the current pixel area to be detected;
[0012] S5. Calculate a position error value according to the reference boundary pixel position and the physical actual opposite edge pixel position, and control the opposite edge mechanism of the spreading machine to move opposite edges according to the position error value.
[0013] As a preferred embodiment of the present invention, the opposite edge area includes an inner texture area, an outer texture area and a fabric-free area.
[0014] As a preferred embodiment of the present invention, S1 further comprises the steps of:
[0015] S11, obtaining image pixel values in the sampling pixel area;
[0016] S12, converting the image pixel values into a gray level co-occurrence matrix, and calculating standard feature quantities.
[0017] As a preferred embodiment of the present invention, both the standard feature quantity and the actual feature quantity include feature quantities in three dimensions: energy, entropy and correlation.
[0018] As a preferred embodiment of the present invention, in S3, the specific calculation method of the feature measurement value is: according to the standard feature quantity of the sampling pixel area and the actual feature quantity of the pixel area to be detected, the standard deviation corresponding to each opposite side area is calculated, and the standard deviation value is the feature measurement value.
[0019] As a preferred embodiment of the present invention, in S3, the specific method of determining the type of the opposite side region into which the currently selected pixel region to be detected falls according to the feature metric value is to compare the size relationship of the feature metric values corresponding to each opposite side region.
[0020] As a preferred embodiment of the present invention, in S4, when selecting the next pixel area to be detected, it moves by a distance of ε pixels, and the pixel distance ε is smaller than the pixel value of the side length of the opposite side area.
[0021] As a preferred embodiment of the present invention, in S5, the position error value is:
[0022] e=k(x * -x)
[0023] Where e is the position error value, x *is the horizontal coordinate value of the reference boundary pixel position, x is the horizontal coordinate value of the actual physical opposite edge pixel position, and k is the calibration gain.
[0024] As a preferred embodiment of the present invention, in S3, the specific calculation method of the feature measurement value is: according to the standard feature quantity of the sampling pixel area and the actual feature quantity of the pixel area to be detected, the standard deviation corresponding to each opposite side area is calculated, and the standard deviation value is the feature measurement value.
[0025] As a preferred embodiment of the present invention, in S3, the specific method of determining the type of the opposite side region into which the currently selected pixel region to be detected falls according to the feature metric value is to compare the size relationship of the feature metric values corresponding to each opposite side region.
[0026] In summary, the present invention has the following beneficial effects:
[0027] 1. In the present invention, by dividing different types of opposite edge regions and using the color processing method of the grayscale co-occurrence matrix, complex textures are converted into a processable form, and feature extraction and contrast calculation are used to determine the location of the pixel area to be detected by judging the numerical value of the feature measurement value, thereby determining the reference boundary pixel position. The predicted position is extremely accurate, and with the help of previous specific texture learning, the accuracy is much higher than that of traditional photoelectric detection methods.
[0028] 2. In addition, by calculating the position error value e and the numerical value of the characteristic measurement value, the movement speed and direction of the servo motor responsible for driving the edge drive mechanism are controlled, which meets the accurate edge requirements and solves the edge problem of composite fabrics. BRIEF DESCRIPTION OF THE DRAWINGS
[0029] In order to more clearly illustrate the technical solutions in the embodiments of this specification, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this specification. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0030] Figure 1 A flowchart of the method is shown in FIG.
[0031] Figure 2 This is a schematic diagram of dividing the opposite edge areas of the fabric in this embodiment;
[0032] Figure 3 This is a schematic diagram of the edge alignment process in this embodiment;
[0033] Figure 4 Schematic diagram of the process in this embodiment. DETAILED DESCRIPTION
[0034] The technical solutions of the embodiments of the present invention are explained and described below in conjunction with the drawings of the embodiments of the present invention, but the following embodiments are only preferred embodiments of the present invention, not all. Based on the embodiments in the implementation mode, other embodiments obtained by those skilled in the art without creative work are all within the protection scope of the present invention.
[0035] The terms "first", "second", etc. in the description and claims of this specification and the above-mentioned drawings are used to distinguish different objects, rather than to describe a specific order. In addition, the term "comprising" and any variation thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device comprising a series of steps or units is not limited to the listed steps or units, but optionally also includes steps or units that are not listed, or optionally also includes other steps or units inherent to these processes, methods, products or devices.
[0036] Therefore, if Figure 1 As shown, before starting to lay the cloth, it is necessary to first divide the opposite edge areas. Compared with the type of fabric to be laid to be solved by the present invention, it mainly includes three types of opposite edge areas, namely the inner texture area occupied by the fabric body, the outer texture area of the outer hair fringes of the fabric body and the outermost cloth-free area.
[0037] Among them, in the present invention, Figure 2 As shown, the opposite edge is the boundary edge between the inner texture area and the outer texture area. The purpose of the present invention is to accurately find the opposite edge through visual alignment and control the opposite drive mechanism of the spreading machine to drive the fabric to always move in the correct opposite direction.
[0038] Then, if Figure 3 and Figure 4 As shown, the texture learning step before spreading the fabric is performed. The inner texture area, the outer texture area and the non-fabric area of the fabric are aligned with the industrial camera lens by a human or machine, and the range of the sampling pixel area is selected on the operation interface, and a sampling pixel area is obtained in each opposite side area. Specifically, the industrial camera reads the pixel values of the above sampling pixel areas respectively, which can be N×N pixels. When the fabric is colored, that is, the pixel value is in RGB format, the grayscale value can be compressed from 255 levels to N (N<255) levels by linear transformation.
[0039] After that, it is necessary to calculate the standard feature quantity of each sampling pixel area; before calculating the standard feature quantity, it is necessary to first classify the features of the above sampling pixel area, specifically, convert the pixel values of the above sampling pixel area into a gray level co-occurrence matrix, and use the K neighboring values of the gray level co-occurrence matrix as the judgment criterion. Obtain the image pixel values in the sampling pixel area, and then calculate the gray level co-occurrence matrix P of the inner texture area, outer texture area and no cloth area respectively. δ(i, j), where δ = (1, 0) or (0, 1) or (1, 1) or (1, -1), represents the displacement of each pixel in the region along the x-axis coordinate and the y-axis coordinate pixel, that is, the first pixel to the right, the first pixel directly above, the first pixel above the right, and the first pixel above the left of each pixel. (i, j) represents the index of the grayscale level of the two pixels, P δ The elements in (i, j) represent the number or probability that the grayscale levels of a pair of pixels defined by δ in the analysis area of the image meet the combination of (i, j). Four grayscale co-occurrence matrices can be calculated for the inner texture area, the outer texture area and the non-texture area respectively;
[0040] The inner texture gray level co-occurrence matrix is denoted as P (1，0) (i, j), P (0，1) (i, j), P (1，1) (i, j), P (1，-1) (i, j);
[0041] The gray-level co-occurrence matrix of the outer texture is recorded as
[0042] The gray-level co-occurrence matrix of the non-clothed area is denoted as P k (1,0) (i,j),P k (0，1) (i,),P k (1，1) (i, j), P k (1，-1) (i, j); then, according to the calculated gray level co-occurrence matrix, the following standard feature quantities are calculated respectively:
[0043]
[0044] Among them, E δ is the standard energy characteristic of the inner texture area, H δ is the standard entropy characteristic of the inner texture area, C δ is the standard correlation feature of the inner texture area, is the standard energy characteristic of the outer texture area, is the standard entropy feature of the outer texture area, is the standard correlation feature of the outer texture area, is the standard energy characteristic quantity of the no-distribution area, is the standard entropy characteristic of the no-distribution area, is the standard correlation characteristic quantity of the no-distribution area, where:
[0045]
[0046] Next, we start spreading the cloth, select a pixel area to be detected (also N×N pixel area) and calculate its actual feature quantity: E δ is the actual energy characteristic of the inner texture area, H δ is the actual entropy feature of the inner texture area, C δ is the actual correlation feature of the inner texture area, is the actual energy characteristic of the outer texture area, is the actual entropy feature of the outer texture area, is the actual correlation feature of the outer texture area, is the actual energy characteristic quantity of the unpopulated area, is the actual entropy characteristic of the no-distribution area, is the actual correlation feature quantity of the non-distribution area. This process is the same as the above process.
[0047] Next, a feature measurement value is calculated based on the standard feature quantity and the actual feature quantity. The specific calculation method of the feature measurement value is: based on the standard feature quantity of the sampling pixel area and the actual feature quantity of the pixel area to be detected, the standard deviation corresponding to each pair of edge areas is calculated, and the standard deviation value is the feature measurement value.
[0048]
[0049] Among them, s is the characteristic metric value of the inner texture area, s * is the characteristic metric value of the outer texture area, s k is the characteristic metric value of the no-cloth area.
[0050] If k <s,s * , indicating that the pixel area to be detected is in a non-clothed area. Continue scanning to the right. When the next pixel area to be detected is selected, move ε pixels away, and the pixel distance ε is less than the side length pixel value of the opposite side area.
[0051] When * <s k ,s, indicating that the feature metric of the pixel area to be detected is closer to the outside, so it can be judged that the area is in the outer texture area, and the calculation continues to the next area to the right until the feature metric of the scanned area satisfies s<s * , it can be determined that the area is in the inner texture area. At this time, the reference boundary pixel position of the current pixel area to be detected is output; wherein the reference boundary pixel position is a pixel side in the opposite direction of the selected direction of the pixel area to be detected in each pixel area to be detected. This reference boundary pixel position is the visual prediction result of the opposite edge position in the present invention.
[0052] In each control cycle, after completing the above image detection, it can be determined whether the opposite edge position is within the image detection range, and according to s, s k ,s * The value of controls the speed and direction of the servo motor responsible for driving the edge-contacting mechanism, satisfying the following logic:
[0053] (1) If the reference boundary pixel position of the current pixel area to be detected is detected, the position error value e is calculated.
[0054] e=k(x * -x)
[0055] Where e is the position error value, x * is the horizontal coordinate value of the reference boundary pixel position, x is the horizontal coordinate value of the actual physical opposite edge pixel position, and k is the calibration gain.
[0056] Then, the position error value is calculated by comparison to obtain a reliable feedback control signal to control the movement speed and direction of the servo motor responsible for driving the edge-aligning drive mechanism, thereby meeting the accurate edge-aligning requirements and solving the edge-aligning problem of composite fabrics.
[0057] (2) If the reference boundary pixel position is not detected, the speed of the servo motor responsible for driving the edge-to-edge mechanism will be constant, but the direction is determined by the current position of the pixel area to be detected. The direction is determined by s, s k ,s * The value of is determined by k <s or s * <s, indicating that the current pixel area to be detected is not located in the inner texture area, and the given direction will be consistent with the direction of the non-textured area; conversely, it indicates that the current pixel area to be detected is all located in the inner texture area, and the given direction will be consistent with the direction of the inner texture area, so as to ensure that the opposite side position always moves toward the pixel area to be detected.
[0058] In another possible embodiment, for texture features, the present invention adopts gray-level co-occurrence matrix and its derived feature quantities to describe them, and may also adopt other methods based on texture structure and statistics to describe them; in the texture classification method, the present invention adopts k-nearest neighbor method, and may also adopt support vector machine or neural network classification methods; in the edge driving method, the present invention adopts servo motor driving method, and may also apply frequency conversion or hydraulic driving methods.
[0059] The above descriptions are only preferred embodiments disclosed in this application and descriptions of the technical principles used. Those skilled in the art should understand that the scope of protection involved in this disclosure is not limited to the technical solutions formed by a specific combination of the above technical features, but should also cover other technical solutions formed by any combination of the above technical features or their equivalent features without departing from the above disclosed concept. For example, the above features are replaced with the technical features with similar functions disclosed in this disclosure (but not limited to) to form a technical solution.
[0060] In addition, although several specific implementation details are included in the above discussion, these should not be interpreted as limiting the scope of the present disclosure. Certain features described in the context of a separate embodiment can also be implemented in a single embodiment in combination. On the contrary, the various features described in the context of a single embodiment can also be implemented in multiple embodiments individually or in any suitable sub-combination.
Claims
1. A visual edge control method for a spreading machine for complex textures and composite fabrics, characterized in that: It includes the following steps: S1, dividing the opposite edge regions, obtaining a sampling pixel region in each opposite edge region, and calculating the standard feature quantity of each sampling pixel region; S2, start spreading, select a pixel area to be detected and calculate its actual feature quantity; S3, calculating a feature measurement value according to the standard feature quantity and the actual feature quantity; judging the type of the opposite side region into which the currently selected pixel region to be detected falls according to the feature measurement value; S4, if it falls into the non-target opposite edge area, automatically select the next pixel area to be detected along the arrangement direction of the opposite edge area and judge again until the selected pixel area to be detected falls into the target opposite edge area, and output the reference boundary pixel position of the current pixel area to be detected; S5. Calculate a position error value according to the reference boundary pixel position and the physical actual opposite edge pixel position, and control the opposite edge mechanism of the spreading machine to move opposite edges according to the position error value.
2. According to claim 1, a visual edge control method for a spreading machine for complex textures and composite fabrics is characterized in that: The opposite edge area includes an inner texture area, an outer texture area and a fabric-free area.
3. The visual edge control method for a spreading machine for complex textures and composite fabrics according to claim 2 is characterized in that: S1 also includes the following steps: S11, obtaining image pixel values in the sampling pixel area; S12, converting the image pixel values into a gray level co-occurrence matrix, and calculating standard feature quantities.
4. The visual edge control method for a spreading machine for complex textures and composite fabrics according to claim 3 is characterized in that: The standard feature quantity and the actual feature quantity both contain feature quantities in three dimensions: energy, entropy and correlation.
5. The visual edge control method of a spreading machine for complex texture and composite fabrics according to claim 4 is characterized in that: In S3, the specific calculation method of the feature measurement value is: according to the standard feature quantity of the sampling pixel area and the actual feature quantity of the pixel area to be detected, the standard deviation corresponding to each pair of edge areas is calculated, and the standard deviation value is the feature measurement value.
6. The visual edge control method for a spreading machine for complex texture and composite fabrics according to claim 5 is characterized in that: In S3, a specific method of determining the type of the opposite side region into which the currently selected pixel region to be detected falls according to the feature metric value is to compare the magnitude relationship of the feature metric values corresponding to each opposite side region.
7. The visual edge control method for a spreading machine for complex texture and composite fabrics according to claim 6 is characterized in that: In S4, when the next pixel area to be detected is selected, it is moved by a distance of ε pixels, and the pixel distance ε is smaller than the pixel value of the side length of the opposite side area.
8. The visual edge control method for a spreading machine for complex texture and composite fabrics according to claim 7 is characterized in that: In S5, the position error value is: e=k(x * -x) Where e is the position error value, x * is the horizontal coordinate value of the reference boundary pixel position, x is the horizontal coordinate value of the physical actual opposite edge pixel position, and k is the calibration gain.
9. The visual edge control method for a spreading machine for complex texture and composite fabrics according to claim 1 is characterized in that: The type of the opposite side region into which the pixel region to be detected falls is determined by a k-nearest neighbor method, a support vector machine determination method or a neural network determination method.
10. The visual edge control method for a spreading machine for complex texture and composite fabrics according to claim 1, characterized in that: The driving mode of the edge-contrasting mechanism is servo motor driving, variable frequency motor driving or hydraulic driving.