Vision positioning double-head asynchronous shortest path planning method for underwear cutting machine

Abstract

The application discloses a kind of underwear cutting machine with visual positioning double-head asynchronous shortest path planning method, the application relates to underwear cutting technical field, solve the problem that there is still defect in the positioning accuracy, segmented scientificity, double-head collaboration efficiency of path planning technology, the present application is aimed at the smooth cutting section and fine cutting section after division, respectively using "the shortest invalid path" strategy optimal path planning: when planning, by "near endpoint connection", reduce the cutting head moving empty course, reduce invalid driving time;At the same time, two groups of cutting heads undertake the cutting task of smooth cutting section and fine cutting section, combined with asynchronous collaboration mode, without waiting for each other, and through time difference optimal division strategy reduces the task time difference of two groups of cutting heads, realizes load balancing, greatly improves the synergy and overall processing efficiency of double-head collaboration.

Description

Technical Field

[0001] This invention relates to the field of underwear cutting technology, specifically to a visual positioning dual-head asynchronous shortest path planning method for underwear cutting machines. Background Technology

[0002] Due to the diverse styles and the prevalence of soft fabrics such as lace and knitted materials, the cutting and processing of lingerie products demands high precision, efficiency, and material utilization. Currently, the lingerie cutting industry largely utilizes automated cutting machines combined with path planning technology to replace the low efficiency and precision of traditional manual cutting. However, existing technologies still face several challenges in practical application that urgently need to be addressed. Existing technologies often employ a uniform cutting rate or a simple division based on graphic regions. Some solutions fail to subdivide the curvature variations of the cutting contour, uniformly using either high-speed or low-speed cutting: at high speeds, delicate curved sections (such as lace cutouts in underwear, shoulder strap joints) are prone to edge burrs and pattern deformation; at low speeds, smooth curved sections (such as the large-radius contours of bra cups) have extremely low processing efficiency, making it impossible to balance quality and efficiency. Another solution, while segmenting, lacks scientific rigor in its segmentation criteria and fails to optimize the combination of cutting rate and task duration, leading to load imbalance during subsequent dual-head collaboration. Regarding the collaboration and path optimization of dual-head cutting machines, existing technologies mostly adopt a synchronous collaboration mode, meaning the two cutting heads need to maintain coordinated movements. This easily leads to an efficiency bottleneck where the "high-speed cutting head waits for the low-speed cutting head." Furthermore, in the path planning of multi-segment cutting trajectories, the optimization of the cutting head's idle travel is not fully considered, often employing a sequential connection method, resulting in excessively long ineffective movement paths and further reducing processing efficiency. In addition, existing path planning schemes are not fully adapted to the flexible fabric characteristics of underwear cutting; the smoothness of the cutting head movement is insufficient, easily causing fabric displacement due to vibration, affecting cutting accuracy. In summary, current path planning technology for underwear cutting machines still has significant shortcomings in terms of positioning accuracy, scientific segmentation, efficiency of dual-head collaboration, and material utilization, making it difficult to meet the needs of refined and efficient underwear production. Summary of the Invention

[0003] To address the shortcomings of existing technologies, this invention provides a visual positioning dual-head asynchronous shortest path planning method for underwear cutting machines, which solves the problems of deficiencies in path planning technology in terms of positioning accuracy, segmentation scientificity, and dual-head collaboration efficiency.

[0004] To achieve the above objectives, the present invention provides the following technical solution: a visual positioning dual-head asynchronous shortest path planning method for an underwear cutting machine. Step 1: Use machine vision equipment to acquire surface images of the underwear fabric, convert the surface images to grayscale, then use the Sobel algorithm to identify the built-in contour partitions, and mark the partitions to be cut using a preset standard template. Confirm the center point A of the built-in contour partition, simultaneously identify the center point B of the standard template, make center point A coincide with center point B, and control the standard template to rotate and scale according to center point B. Record whether the built-in contour partition and the standard template completely coincide in several processing processes. If they do, mark the built-in contour partition as the partition to be cut. If they do not, do not perform any marking process. Step 2: Perform smoothness analysis on the identified partitions to be cut. By combining a two-dimensional coordinate system, confirm the curvature associated with different contour points on the contour of the partition to be cut. Based on the curvature variation characteristics between adjacent contour points, divide the partition contour of the partition to be cut into multiple segments with similar curvature. The specific method is as follows: The curvature K of adjacent contour points is successively adjusted. j Perform difference verification and process the absolute value of the confirmed difference to obtain the value to be verified. If the value to be verified is less than or equal to Y2, then the adjacent contour points are recorded as contour points of the same type. Otherwise, no marking is performed. Y2 is a preset value. The contour line segments associated with the same type of contour points are recorded as similar curvature segments, and several sets of similar curvature segments existing in the partition contour are confirmed in turn. Step 3: Perform feature classification on the multiple similar curvature segments, sequentially select boundary values ​​to divide the multiple similar curvature segments, and lock the association time in combination with the preset cutting rate. From the association times corresponding to multiple selected processes, select the optimal process and complete the segmentation process of smooth cutting segments or fine cutting segments. The curvature of several contour points within a single set of similar curvature segments is averaged and recorded as the segmentation feature of the corresponding similar curvature segment. The segmentation features of several sets of similar curvature segments are sorted in ascending order of value to confirm the segmentation feature value column. Within the numerical range associated with the segmented feature value column, randomly select a boundary value: Segments with similar curvature whose segmentation feature is less than or equal to the boundary value are denoted as smooth cut segments, and segments with similar curvature whose segmentation feature is greater than or equal to the boundary value are denoted as fine cut segments. The bus length of the smooth cut segment corresponding to the currently selected process is denoted as LP. k The bus length of the finely segmented section is denoted as LZ. k , where k represents different selection processes; Extract the preset cutting rate V1 for the smooth cutting segment and the preset cutting rate V2 for the fine cutting segment, where V1 > V2, and use: T1 k =LP k÷V1 and T2 k =LZ k ÷V2 confirms the characteristic time T1 associated with the two sets of cut segments. k and T2 k And adopt: CZ k =|T1 k -T2 k |Confirmation time difference CZ k The time difference is used as the associated time for the selected process. From the different associated times of different selection processes, the minimum value is selected, the selection process corresponding to the minimum value is recorded as the optimal process, and the smooth cutting segment or fine cutting segment divided by the optimal process is extracted and recorded. Step 4: Based on the divided smooth or fine cutting segments, perform path planning for different types of cutting segments, select the optimal path from multiple planning processes, and execute it. For the marked smooth cutting segments, identify the total number G of smooth cutting segments. If G is 1, randomly select one endpoint of the smooth cutting segment as the starting point and the other endpoint as the ending point, generate a cutting path and execute it as the optimal path. If G≥2, a smooth cutting segment is randomly selected as the starting segment. Any endpoint of the starting segment is used as the starting point, and the first set of cutting paths is generated. Then, the endpoint of the starting segment is used as the reference point. The endpoint closest to the reference point is identified from the different endpoints of other smooth cutting segments, and this endpoint is recorded as the second starting point. The second set of cutting paths is generated, and so on, until all smooth cutting segments have been selected. Invalid paths are extracted. Invalid paths are not cutting paths but moving paths. The length of invalid paths is recorded as L. Several selection processes are executed. The selection process with the minimum value of invalid path L is taken as the optimal planning process. The multiple sets of cutting paths identified by the optimal planning path are integrated into a whole cutting path and executed as the optimal path. For the fine cutting segment, the same confirmation method as the smooth cutting segment is adopted to lock the optimal path, and the two locked optimal paths are cut by two sets of cutting heads in sequence. The cutting rate of the optimal path of the smooth cutting segment is V1, and the cutting rate of the optimal path of the fine cutting segment is V2.

[0005] This invention provides a vision-based, dual-head asynchronous shortest path planning method for an underwear cutting machine. Compared with existing technologies, it has the following advantages: By calculating the curvature and verifying the difference between adjacent curvatures, the partition contour is divided into segments with similar curvatures. Then, by combining the preset cutting rate value and the optimal time difference strategy, smooth cutting segments and fine cutting segments are divided. This allows high-speed cutting (V1) to be adapted to smooth curve segments and low-speed fine cutting (V2) to be adapted to complex curve segments. This avoids the waste of efficiency when cutting smooth curve segments and prevents edge burrs and pattern deformation problems caused by high-speed cutting in fine curve segments. It is especially suitable for cutting scenarios with different precision requirements such as bra cup curvature and lace cutouts. For the divided smooth cutting segments and fine cutting segments, the "shortest invalid path" strategy is adopted to plan the optimal path: when planning multiple cutting segments of the same type, the "connection of nearest endpoints" reduces the idle travel of the cutting head and reduces the invalid travel time; at the same time, the two sets of cutting heads respectively undertake the cutting tasks of the smooth cutting segment and the fine cutting segment. Combined with the asynchronous cooperation mode, there is no need for mutual waiting. Moreover, the task time difference between the two sets of cutting heads is reduced by the optimal division strategy of time difference, so as to achieve load balance and greatly improve the coordination of dual-head cooperation and the overall processing efficiency. In the process of identifying the areas to be cut, the system adapts to the randomly distributed areas on the fabric through flexible rotation and scaling comparison, eliminating the need for pre-processing the fabric to be neatly sorted and making full use of the effective area of ​​the fabric. At the same time, accurate area identification avoids the increase of cutting waste caused by positioning deviation, improving the fabric utilization rate from the source, reducing material waste, and lowering the material cost of underwear production. Attached Figure Description

[0006] Figure 1 This is a schematic diagram of the method flow of the present invention. Detailed Implementation

[0007] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0008] First Embodiment Please see Figure 1 This application provides a visual positioning dual-head asynchronous shortest path planning method for an underwear cutting machine, including the following steps: Step 1: Use machine vision equipment to acquire surface images of the underwear fabric, and process the surface images into grayscale. Then, use the Sobel algorithm to confirm the built-in contour partitions. Combined with the preset standard template, mark the partitions to be cut. Specifically, the fabric passes through the corresponding cutting instrument in sequence through the corresponding conveyor device. The vision detection device above it acquires images of the specified area, processes them into grayscale, and identifies the partition contours in the corresponding area based on the grayscale features associated with different pixels. Then, combined with the corresponding standard template, lock the relevant partitions on the surface that need to be cut. Step 2: Perform smoothness analysis on the identified partitions to be cut. By combining the two-dimensional coordinate system, confirm the curvature associated with different contour points on the contour of the partition to be cut. Based on the curvature change characteristics between adjacent contour points, divide the partition contour of the partition to be cut into multiple similar curvature segments. Specifically, in order for the two cutting heads to perform different processing processes, it is necessary to effectively divide the corresponding partition contours into smooth curve segments with faster cutting rates and fine curve segments with slower cutting rates. Therefore, it is necessary to confirm the relevant similar curvature segments and perform comprehensive division. Step 3: Perform feature classification on the multiple similar curvature segments, select the boundary value in turn to divide the multiple similar curvature segments, and lock the association time in combination with the preset cutting rate. Select the optimal process from the association time corresponding to multiple selected processes, and complete the segmentation process of smooth cutting segment or fine cutting segment. Step 4: Based on the divided smooth or fine cutting segments, perform path planning for different types of cutting segments, select the optimal path from multiple planning processes, and execute it.

[0009] Second Embodiment In the specific implementation process of this embodiment, the focus is mainly on the specific marking process of the corresponding partition to be cut in step one; The specific marking process for the partition to be cut includes: The surface image of the underwear fabric in the monitoring area is converted to grayscale. The different RGB values ​​associated with different pixels in the surface image are identified. Different weights are then assigned to the RGB values ​​to confirm the grayscale value associated with the corresponding pixel. Grayscale value = R×0.114+G×0.587+B×0.299. Based on the different grayscale values ​​associated with different pixels, the corresponding surface image is converted to grayscale to confirm the grayscale image. The Sobel algorithm is used to confirm the vertical and horizontal gradients associated with different grayscale points within a grayscale image. (Based on other grayscale points surrounding the corresponding grayscale point, the different grayscale values ​​associated with that point are confirmed. Combined with the corresponding weighting factors, the vertical and horizontal gradients associated with the corresponding grayscale point can be effectively confirmed. Subsequently, the overall gradient associated with the corresponding grayscale point can be effectively confirmed. Since the Sobel algorithm for confirming gradient pixels is a common technique in existing technologies, it will not be elaborated upon here.) Identify the comprehensive gradient associated with each grayscale point and record grayscale points that satisfy the condition: comprehensive gradient ≥ Y1 as gradient pixels; otherwise, no marking is made. Y1 is a preset value determined in advance by the operator based on the fabric's weave depth. Specific cutting lines are pre-marked on the fabric, and the identified gradient pixels are the points on these lines. For cotton underwear fabric with a weave depth H = 0.3 mm, if calibration experiments yield k1 = 100 and b1 = 30, then Y1 = 100 × 0.3 + 30 = 6. 0; At this point, grayscale points with a comprehensive gradient ≥ 60 will be marked as gradient pixels, accurately corresponding to the segmentation line position. Through pre-conducted sample calibration experiments, an empirical correlation formula between the crochet depth H and the gradient threshold Y1 is established: Y1 = k1 × H + b1, where k1 is the gradient sensitivity coefficient (unit: gradient value / mm), which is determined by the fabric material (e.g., k1 is 80~120 for cotton and 150~200 for spandex blends), and b1 is the basic gradient threshold (unit: gradient value), used to compensate for the uniformity deviation of the fabric's base grayscale, with a value range of 20~50; Several groups of consecutive gradient pixels are connected sequentially to generate gradient contour lines. The area enclosed by the closed gradient contour lines is denoted as the built-in contour partition. The confirmed sets of built-in contour partitions are then compared and verified with the standard template. Confirm the center point A of the built-in contour partition (by combining the two-dimensional coordinate system, confirm the two-dimensional coordinates of different contour points, and then perform averaging to directly confirm its center point), simultaneously identify the center point B of the standard template (a preset template), make center point A coincide with center point B, and control the standard template to rotate and scale according to center point B. Record whether there is a state in several processing processes where the built-in contour partition and the standard template completely coincide. If so, mark this built-in contour partition as a partition to be cut. If not, do not perform any marking process. Specifically, to make full use of the fabric, the corresponding area of ​​the fabric is fully utilized when marking lines. Therefore, each section to be cut is not ordered in a regular way, but randomly. Some are ordered in the forward direction and some in the reverse direction. In order to effectively identify each segmented area, the segmented area needs to be rotated and scaled with the corresponding standard template at the center point. This allows the corresponding segmented area to be aligned with the section to be cut for verification, thus enabling quick and effective identification of the section to be cut. Each identified section to be cut is basically the same.

[0010] Third Embodiment In the specific implementation process, compared with the above embodiments, this embodiment mainly focuses on the comprehensive division and processing of similar curvature segments; The specific method for confirming the curvature corresponding to different contour points on the contour of the partition to be segmented is as follows: Based on the overlap verification process between the partition to be segmented and the standard template (in the state of complete overlap), the preset cutting contour in the standard template is synchronously marked in the partition to be segmented, and the cutting contour marked in the partition to be segmented is recorded as the partition contour. Construct a set of two-dimensional coordinate systems, with the origin of the two-dimensional coordinate systems coinciding with the center point of the partition to be cut, and sequentially record the contour coordinates (X, Y, X) associated with different contour points within the partition contour. i Y i ), where i represents different contour points. Prioritize determining the first derivative coordinates of the corresponding contour points: designate the selected contour point as the principal point, and designate the two adjacent contour points before and after it as secondary points. Mark the contour coordinates of the principal point as (X... j Y j ), where j∈i, and the contour coordinates of the two sub-points are denoted as (X j+1 Y j+1 ) and (X j-1 Y j-1 ),use: Confirm the first derivative coordinates (X1) associated with the current principal point. j Y1 j ), where t is a preset value, ranging from 0.01 to 0.05 seconds, and the first derivative coordinates (X1) of the two secondary points are determined in the same way. j+1 Y1 j+1 ) and (X1 j-1 Y1 j-1 Then confirm the second derivative coordinates with respect to the principal point: Based on the first derivative coordinates (X1) determined by the principal point j Y1 j ) and second derivative coordinates (X2) j Y2 j ),use: Confirm the curvature K associated with the principal point j Using the same confirmation method, the curvature of each contour point is calculated (because the production process and the dividing lines are all closed, there are associated contour points before and after any contour point). Here, the quantization derivative of distance and time is used to confirm the corresponding rate characteristics. The specific method for dividing the partition outline to be cut into multiple segments with similar curvature is as follows: The curvature K of adjacent contour points is successively adjusted. j A difference check is performed, and the confirmed difference is processed into an absolute value to obtain the value to be checked (that is, the curvature of adjacent contour points is checked by difference and then processed into an absolute value to obtain the corresponding value to be checked). If the value to be checked is ≤ Y2, the adjacent contour points are recorded as contour points of the same type; otherwise, no marking is made. Y2 is a preset value, and its specific value is confirmed in conjunction with the coordinate accuracy of the coordinate system and is determined in advance by the operator. Coordinate accuracy refers to the smallest resolvable unit (denoted as Δd, unit: pixel) of the contour point coordinates in the two-dimensional coordinate system, which determines the measurement error range of the contour point coordinates. The curvature difference between adjacent contour points (value to be checked) includes two parts: the true curvature change and the false curvature change caused by the coordinate measurement error. The core function of Y2 is to distinguish contour points of the same type (small true curvature change) from contour points of different types (large true curvature change). Its quantitative essence is to convert the coordinate accuracy into the error threshold of the curvature difference, ensuring that Y2 is greater than the maximum value of the false curvature change and avoiding misclassification. The contour segments associated with the same contour points are recorded as similar curvature segments, and several sets of similar curvature segments existing in the partition contour are confirmed in turn (generally there will be no single point omission, because the overall contour of the underwear is marked in a relatively smooth way). Specifically, in order to better complete the segmentation process of several similar curvature segments, the curvature associated with each contour point is comprehensively verified, and the related curvatures with small numerical differences are comprehensively classified into numerical segments. This can effectively divide the cutting contour of the corresponding partition into multiple related segments with similar curvatures, which facilitates the subsequent comprehensive evaluation of curvature segments and the path planning of the corresponding cutting head.

[0011] Third Embodiment In this embodiment, compared to the above embodiments, the specific implementation process mainly focuses on the segmentation process of the smooth cutting segment or fine cutting segment in step three. The specific segmentation process includes: The curvature of several contour points within a single set of similar curvature segments is averaged and recorded as the segmentation feature of the corresponding similar curvature segment. The segmentation features of several sets of similar curvature segments are sorted in ascending order of value to confirm the segmentation feature value column. Within the numerical range associated with the segmented feature value column, randomly select a boundary value: Segments with similar curvature whose segmentation feature is less than or equal to the boundary value are denoted as smooth cut segments, and segments with similar curvature whose segmentation feature is greater than or equal to the boundary value are denoted as fine cut segments. The bus length of the smooth cut segment corresponding to the currently selected process is denoted as LP. k The bus length of the finely segmented section is denoted as LZ. k , where k represents different selection processes; Extract the preset cutting rate V1 for the smooth cutting segment and the preset cutting rate V2 for the fine cutting segment, where V1 > V2, and use: T1 k =LP k ÷V1 and T2 k =LZ k ÷V2 confirms the characteristic time T1 associated with the two sets of cut segments. k and T2 k And adopt: CZ k =|T1 k -T2 k |Confirmation time difference CZ k The time difference is used as the associated time for the selected process. From the different associated times of different selection processes, the minimum value is selected, the selection process corresponding to the minimum value is recorded as the optimal process, and the smooth cutting segment or fine cutting segment divided by the optimal process is extracted and recorded. Specifically, in order for both sets of cutting heads to complete the cutting of a single segmentation area quickly and effectively, it is necessary to combine the preset parameter rate and the corresponding line length to quickly and effectively identify the characteristic time associated with the corresponding travel process, and verify the process based on the difference of the characteristic time, and comprehensively divide the smooth cutting segment and the fine cutting segment to facilitate the subsequent processing of the corresponding cutting head path planning.

[0012] The process of selecting the optimal path in step four specifically includes: For the marked smooth cutting segments, identify the total number G of smooth cutting segments. If G is 1, randomly select one endpoint of the smooth cutting segment as the starting point and the other endpoint as the ending point, generate a cutting path and execute it as the optimal path. If G≥2, a smooth cutting segment is randomly selected as the starting segment. Any endpoint of the starting segment is used as the starting point, and the first set of cutting paths is generated. Then, the endpoint of the starting segment is used as the reference point. The endpoint closest to the reference point is identified from different endpoints of other smooth cutting segments, and this endpoint is recorded as the second starting point. The second set of cutting paths is generated, and so on, until all smooth cutting segments have been selected. Invalid paths are extracted. Invalid paths are not cutting paths, but moving paths (moving the end point of the starting segment to the starting point of the next smooth cutting segment). The length of the invalid path is recorded as L. Several selection processes are executed. The selection process with the minimum value of invalid path L is taken as the optimal planning process. The multiple sets of cutting paths identified by the optimal planning path are integrated into a whole cutting path and executed as the optimal path. For the fine cutting segment, the optimal path is confirmed to be the same as that of the smooth cutting segment. The optimal path is locked and the two locked optimal paths are cut by two sets of cutting heads in sequence. The cutting rate of the optimal path of the smooth cutting segment is V1 and the cutting rate of the optimal path of the fine cutting segment is V2. Specifically, in order to achieve better cutting and planning results, it is necessary to process the relevant paths of different road segments, select the optimal path, so that each cutting head can effectively reduce the ineffective travel time, and at the same time, fully improve its processing effect, reduce the time difference between different cutting heads, and improve the cutting effect of dual-head collaboration.

[0013] Some of the data in the above formulas are numerical calculations with dimensions removed, and the contents not described in detail in this specification are all prior art known to those skilled in the art.

[0014] The above embodiments are only used to illustrate the technical methods of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical methods of the present invention without departing from the spirit and scope of the technical methods of the present invention.

Claims

1. A visual positioning dual-head asynchronous shortest path planning method for an underwear cutting machine, characterized in that, Includes the following steps: Step 1: Use machine vision equipment to acquire surface images of the underwear fabric, process the surface images into grayscale, then use the Sobel algorithm to confirm the built-in contour partitions, and mark the partitions to be cut by combining them with the preset standard templates. Step 2: Perform smoothness analysis on the identified partitions to be cut. By combining the two-dimensional coordinate system, confirm the curvature associated with different contour points on the contour of the partition to be cut. Based on the curvature change characteristics between adjacent contour points, divide the partition contour of the partition to be cut into multiple segments with similar curvature. Step 3: Perform feature classification on the multiple similar curvature segments, sequentially select boundary values ​​to divide the multiple similar curvature segments, and lock the association time in combination with the preset cutting rate. From the association times corresponding to multiple selected processes, select the optimal process and complete the segmentation process of smooth cutting segments or fine cutting segments. Specifically, this includes: The curvature of several contour points within a single set of similar curvature segments is averaged and recorded as the segmentation feature of the corresponding similar curvature segment. The segmentation features of several sets of similar curvature segments are sorted in ascending order of value to confirm the segmentation feature value column. Within the numerical range associated with the segmented feature value column, randomly select a boundary value: Segments with similar curvature whose segmentation feature is less than or equal to the boundary value are denoted as smooth cut segments, and segments with similar curvature whose segmentation feature is greater than or equal to the boundary value are denoted as fine cut segments. The bus length of the smooth cut segment corresponding to the currently selected process is denoted as LP. k The bus length of the finely segmented section is denoted as LZ. k , where k represents different selection processes; Extract the preset cutting rate V1 for the smooth cutting segment and the preset cutting rate V2 for the fine cutting segment, where V1 > V2, and use: T1 k =LP k ÷V1 and T2 k =LZ k ÷V2 confirms the characteristic time T1 associated with the two sets of cut segments. k and T2 k And adopt: CZ k =|T1 k -T2 k |Confirmation time difference CZ k The time difference is used as the associated time for the selected process. From the different associated times of different selection processes, the minimum value is selected, the selection process corresponding to the minimum value is recorded as the optimal process, and the smooth cutting segment or fine cutting segment divided by the optimal process is extracted and recorded. Step 4: Based on the divided smooth or fine cutting segments, perform path planning for different types of cutting segments, select the optimal path from multiple planning processes, and execute it.

2. The visual positioning dual-head asynchronous shortest path planning method for an underwear cutting machine according to claim 1, characterized in that, In step one, the specific method for marking the partition to be cut is as follows: The surface image of the underwear fabric in the monitoring area is processed into grayscale. The different RGB values ​​associated with different pixels in the surface image are identified. Different weights are assigned to the RGB values ​​to confirm the grayscale value associated with the corresponding pixel. Based on the different grayscale values ​​associated with different pixels, the corresponding surface image is processed into grayscale to confirm the grayscale image. The Sobel algorithm is used to confirm the vertical and horizontal gradients associated with different gray-level points within a grayscale image. Confirm the comprehensive gradient associated with the corresponding grayscale point, and record the grayscale point that satisfies: comprehensive gradient ≥ Y1 as gradient pixel point, otherwise do not make any mark, where Y1 is a preset value; Several groups of consecutive gradient pixels are connected sequentially to generate gradient contour lines. The area included by the closed gradient contour lines is recorded as the built-in contour partition. The confirmed sets of built-in contour partitions are compared and verified with the standard template to lock the partition to be cut.

3. The visual positioning dual-head asynchronous shortest path planning method for an underwear cutting machine according to claim 2, characterized in that, The specific method for comparing and verifying the built-in contour partition with the standard template is as follows: Confirm the center point A of the built-in contour partition, simultaneously identify the center point B of the standard template, make center point A coincide with center point B, and control the standard template to rotate and scale according to center point B. Record whether the built-in contour partition and the standard template completely coincide in several processing processes. If they do, mark the built-in contour partition as the partition to be cut. If they do not, do not perform any marking process.

4. The visual positioning dual-head asynchronous shortest path planning method for an underwear cutting machine according to claim 1, characterized in that, In step two, the specific method for confirming the curvature corresponding to different contour points on the contour to be segmented is as follows: Based on the process of verifying the overlap between the partition to be segmented and the standard template, the preset cutting contours in the standard template are synchronously marked in the partition to be segmented, and the cutting contours marked in the partition to be segmented are recorded as the partition contours. Construct a set of two-dimensional coordinate systems, with the origin of the two-dimensional coordinate systems coinciding with the center point of the partition to be cut, and sequentially record the contour coordinates (X, Y, X) associated with different contour points within the partition contour. i Y i ), where i represents different contour points. Prioritize determining the first derivative coordinates of the corresponding contour points: designate the selected contour point as the principal point, and designate the two adjacent contour points before and after it as secondary points. Mark the contour coordinates of the principal point as (X... j Y j ), where j∈i, and the contour coordinates of the two sub-points are denoted as (X j+1 Y j+1 ) and (X j-1 Y j-1 ),use: Confirm the first derivative coordinates (X1) associated with the current principal point. j Y1 j ), where t is a preset value, and the first derivative coordinates (X1) of the two secondary points are determined in the same way. j+1 Y1 j+1 ) and (X1 j-1 Y1 j-1 Then confirm the second derivative coordinates with respect to the principal point: Based on the first derivative coordinates (X1) determined by the principal point j Y1 j ) and second derivative coordinates (X2) j Y2 j ),use: Confirm the curvature K associated with the principal point j Using the same verification method, the curvature of each contour point is calculated.

5. The visual positioning dual-head asynchronous shortest path planning method for an underwear cutting machine according to claim 4, characterized in that, In step two, the specific method for dividing the partition outline of the area to be cut into multiple segments with similar curvature is as follows: The curvature K of adjacent contour points is successively adjusted. j Perform difference verification and process the absolute value of the confirmed difference to obtain the value to be verified. If the value to be verified is less than or equal to Y2, then the adjacent contour points are recorded as contour points of the same type. Otherwise, no marking is performed. Y2 is a preset value. The contour segments associated with similar contour points are denoted as similar curvature segments, and several sets of similar curvature segments existing within the partition contour are confirmed in sequence.

6. The visual positioning dual-head asynchronous shortest path planning method for an underwear cutting machine according to claim 1, characterized in that, In step four, the specific method for selecting the optimal path is as follows: For the marked smooth cutting segments, identify the total number G of smooth cutting segments. If G is 1, randomly select one endpoint of the smooth cutting segment as the starting point and the other endpoint as the ending point, generate a cutting path and execute it as the optimal path. If G≥2, a smooth cutting segment is randomly selected as the starting segment. Any endpoint of the starting segment is used as the starting point, and the first set of cutting paths is generated. Then, the endpoint of the starting segment is used as the reference point. The endpoint closest to the reference point is identified from the different endpoints of other smooth cutting segments, and this endpoint is recorded as the second starting point. The second set of cutting paths is generated, and so on, until all smooth cutting segments have been selected. Invalid paths are extracted. Invalid paths are not cutting paths but moving paths. The length of invalid paths is recorded as L. Several selection processes are executed. The selection process with the minimum value of invalid path L is taken as the optimal planning process. The multiple sets of cutting paths identified by the optimal planning path are integrated into a whole cutting path and executed as the optimal path. For the fine cutting segment, the same confirmation method as the smooth cutting segment is adopted to lock the optimal path, and the two locked optimal paths are cut by two sets of cutting heads in sequence. The cutting rate of the optimal path of the smooth cutting segment is V1, and the cutting rate of the optimal path of the fine cutting segment is V2.

Images