A trajectory planning method for gluing uppers based on color point cloud data

Abstract

The present invention provides a method for planning the trajectory of glue spraying on the upper based on color point cloud data, which belongs to the field of shoemaking automation, and includes: obtaining the original color point cloud data of the complete upper, and preprocessing the original color point cloud data to obtain the color point cloud data of the upper; dividing the color point cloud data of the upper into two halves, and rotating one of the halves so that it is located in the same reference plane as the other half, and projecting the point cloud in the reference plane to the same plane to obtain a plane image of the upper; processing the plane image of the upper to obtain a specific area and the center line of the specific area; identifying the regional point cloud corresponding to the specific area from the color point cloud data of the upper; obtaining the trajectory points corresponding to the center line of the specific area in the color point cloud data of the upper, and calculating the trajectory posture after curve fitting and smoothing the trajectory points, and obtaining the trajectory of glue spraying on the upper. The present invention can effectively improve the accuracy and efficiency of glue spraying.

Description

Technical Field

[0001] The invention belongs to the field of shoemaking automation, and in particular relates to a shoe upper glue spraying trajectory planning method based on color point cloud data. Background Art

[0002] In the shoe production process, glue spraying is one of the key manufacturing steps. The traditional glue spraying process relies on manual experience or simple mechanical control, which is not only inefficient, but also difficult to ensure the accuracy and repeatability of glue spraying. With the rapid development of computer vision technology, it has become possible to use point cloud data for precise measurement and identification. However, the existing point cloud data processing technology mainly focuses on the geometric shape and spatial position of the upper, while the color information of the upper surface is insufficiently utilized, which is particularly limited in scenarios where different color areas need to be distinguished for glue spraying operations. Summary of the invention

[0003] The purpose of the present invention is to propose a shoe upper glue spraying trajectory planning method based on color point cloud data, which can effectively improve the accuracy and efficiency of glue spraying.

[0004] The present invention is achieved through the following technical solutions:

[0005] A shoe upper glue spraying trajectory planning method based on color point cloud data comprises the following steps:

[0006] Step S1, obtaining original color point cloud data of a complete shoe upper, and preprocessing the original color point cloud data to obtain color point cloud data of the shoe upper;

[0007] Step S2, dividing the upper color point cloud data into two halves, and rotating one of the halves so that it is located in the same reference plane as the other half, and projecting the point cloud in the reference plane to the same plane to obtain a upper plane image;

[0008] Step S3, processing the upper plane image to obtain a specific area and a center line of the specific area;

[0009] Step S4, identifying a regional point cloud corresponding to a specific area from the upper color point cloud data;

[0010] Step S5, obtaining the trajectory points corresponding to the center line of the specific area in the upper color point cloud data, and calculating the trajectory posture after curve fitting and smoothing the trajectory points to obtain the upper glue spraying trajectory.

[0011] Furthermore, in step S1, a 3D color camera is used to perform a 360-degree scan on the shoe upper to obtain original color point cloud data of the complete shoe upper, and the original color point cloud data includes geometric shape information of the shoe upper and RGB color information of the shoe upper surface.

[0012] Furthermore, in the step S1, preprocessing the original color point cloud data specifically includes: using a straight-through filtering method to obtain a preliminary ROI area containing the upper, using a statistical filtering method to remove outliers in the preliminary ROI area and implementing downsampling through a voxelized grid method to obtain the upper color point cloud data A.

[0013] Furthermore, in step S2, the short axis direction of the upper color point cloud data is the x-axis, the long axis direction is the y-axis, and the vertical upward direction is the z-axis. The upper color point cloud data is divided into two halves along the y-axis direction, and one of the halves is rotated 180° around a rotation axis parallel to the z-axis so that the two halves are in the same reference plane, and then the two halves are projected onto the yz plane to obtain the upper plane image.

[0014] Furthermore, the step S2 specifically includes the following steps:

[0015] Step S21, obtaining the maximum value point H of the toe in the y-axis direction of the upper color point cloud data, taking the plane passing through the maximum value point and parallel to the yz plane as the dividing plane, classifying the point cloud with an x ​​coordinate value less than the x coordinate value of the point H in the upper color point cloud data as the point set M, and classifying the point cloud with an x ​​coordinate value greater than the x coordinate value of the point H as the point set N, that is, obtaining two halves formed by the point set M and the point set N;

[0016] Step S22, create a rotation axis passing through point H and parallel to the z-axis, rotate the point set N 180° around the rotation axis to obtain the point set O, and merge the point set M and the point set O to obtain the point set Q;

[0017] Step S23: Project the point set Q onto the yz plane, and scale it according to the required resolution and distribution range of the point set Q to obtain the upper plane image.

[0018] Furthermore, the step S23 specifically includes the following steps:

[0019] Step S231, respectively calculating the minimum value min.y and the maximum value max.y of the point set Q in the y dimension, and the minimum value min.z and the maximum value max.z in the z dimension;

[0020] Step S232, create a two-dimensional color image with a width of col and a height of row, and initialize all pixels of the two-dimensional color image to the background color, wherein the width col and the height row of the two-dimensional color image are calculated according to the formulas row=(max.y-min.y)*scale+2*border and col=(max.z-min.z)*scale+2*border respectively, scale represents the set image scaling ratio, border represents the set image border size, and scale and border are set according to the required resolution;

[0021] Step S233, calculate the corresponding position (x, y) of each point q in the point set Q on the two-dimensional color image according to the formulas y=border+(qy-min.y)*scale and x=border+(qz-min.z)*scale, and set the RGB value of the corresponding position to the RGB value of point q to obtain the upper plane image, wherein qy represents the y coordinate value of point q, and qz represents the z coordinate value of point q.

[0022] Furthermore, the step S3 specifically includes the following steps:

[0023] Step S31, converting the RGB color space of the upper plane image into the HSV color space, and selecting the image of the saturation channel in the HSV color space as the processing image;

[0024] Step S32, performing mean smoothing, gray value scaling, and binarization on the processed image in sequence to obtain a high saturation area, and performing threshold processing on the hue image in the high saturation area to extract an area with a specific hue;

[0025] Step S33, performing a filling operation on all regions of a specific hue to form multiple connected domains, selecting the connected domain with the largest area as the target region, and performing closing operations and opening operations on the target region in sequence to fill small holes and remove small protrusions;

[0026] Step S34, extracting the contour of the target area processed in step S33, smoothing the contour to reduce the sawtooth effect and noise points on the contour, and generating a filling area according to the smoothed contour to obtain a specific area R;

[0027] Step S35, applying a Gaussian line extraction algorithm to process the specific area to obtain a line set, and selecting the longest line from the line set as the center line;

[0028] Step S36: convert the sub-pixel precision contour of the center line into a point set coordinate set S, and sample the point set coordinate set S to obtain a two-dimensional point set data set T that retains the center line features.

[0029] Furthermore, the step S4 specifically includes the following steps:

[0030] Step S41, mapping the points of the point set M and the point set O to the two-dimensional color image respectively, and obtaining the two-dimensional point pixel coordinates corresponding to each point;

[0031] Step S42, counting the points in the point set M whose two-dimensional point pixel coordinates are within the specific area into the point set U, and counting the points in the point set O whose two-dimensional point pixel coordinates are within the specific area into the point set V;

[0032] Step S43, rotating the point set V around the rotation axis in the opposite direction to step S22, rotating 180°, to obtain the point set W;

[0033] Step S44: Merge the point set W with the point set U to generate a regional point cloud X corresponding to the specific area.

[0034] Furthermore, the step S5 specifically includes the following steps:

[0035] Step S51, according to the formula x=0, y=(ty-border) / scale+min.y and z=(tz-border) / scale+min.z, convert the points in the two-dimensional point set data set T into the three-dimensional point set X, where (x, y, z) represents the coordinates of the point x' in the three-dimensional point set X, ty represents the y coordinate value of the point t in the two-dimensional point set data set T, and tz represents the z coordinate value of the point t in the two-dimensional point set data set T;

[0036] Step S52: for a point x' in the three-dimensional point set X, the range between the y coordinate values ​​of the previous point and the next point of the point x' is used as a bounding range, a point cloud set within the bounding range is searched in the point set O, and the average of the x coordinate values ​​of each point in the point cloud set is calculated as a temporary x coordinate value of the point x';

[0037] Step S53, obtaining the nearest neighbor point of point x' in the three-dimensional point set X in the point set O, and using the nearest neighbor point as the target point of point x', thereby obtaining a target point set Y corresponding to each point in the three-dimensional point set X;

[0038] Step S54, record the set of points in the target point set Y whose y value is greater than the y value of point H as point set I, record the set of points whose y value is less than the y value of point H as point set J, rotate point set J around the rotation axis in the opposite direction of step S22, rotate 180°, merge the point set obtained after the rotation with point set I, generate a complete trajectory point cloud set, perform B-spline curve fitting and smoothing on the trajectory point cloud set, and obtain an initial glue spraying trajectory point set K;

[0039] Step S55: Calculate the normal vector of each point k in the initial glue spraying trajectory point set K. The normal vector is used as the X-direction vector of the robot tool. n To point k n+1 The direction of the robot tool is taken as the Y direction vector, the Z direction vector is obtained by using the X direction vector and the Y direction vector, and the X direction vector is corrected by using the Y direction vector and the Z direction vector to obtain the gluing trajectory of the shoe upper.

[0040] The present invention has the following beneficial effects:

[0041] 1. The present invention first obtains the original color point cloud data of the complete upper and preprocesses it to obtain the upper color point cloud data, then divides the upper color point cloud data into two halves, rotates one of the halves so that it is located on the same reference plane as the other half, and projects the point cloud in the reference plane to the same plane to obtain an upper plane image, then processes the upper plane image to obtain a specific area and a center line of the specific area, and then identifies the regional point cloud corresponding to the specific area from the upper color point cloud data, and finally obtains the trajectory points corresponding to the center line of the specific area in the upper color point cloud data, and calculates the trajectory posture after curve fitting and smoothing the trajectory points, and finally obtains the upper glue spraying trajectory. In this process, the color information of the upper is fully utilized to achieve accurate identification of specific color areas and accurate planning of glue spraying trajectories, effectively improve the accuracy and efficiency of glue spraying, reduce dependence on manual experience, and is conducive to realizing automated and intelligent production processes. It can be applicable to uppers of different types and sizes, and has good flexibility and scalability. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] The present invention is further described in detail below with reference to the accompanying drawings.

[0043] Figure 1 It is a flow chart of the present invention.

[0044] Figure 2 Schematic diagram of original color point cloud data of the present invention.

[0045] Figure 3 It is a schematic diagram of the color point cloud data of the upper of the present invention.

[0046] Figure 4 It is a schematic diagram of a plane image of a shoe upper of the present invention.

[0047] Figure 5 It is a schematic diagram of a specific area and its center line of the present invention.

[0048] Figure 6 It is a schematic diagram of a regional point cloud corresponding to a specific area of ​​the present invention.

[0049] Figure 7 It is a schematic diagram of the gluing trajectory of the shoe upper of the present invention. DETAILED DESCRIPTION

[0050] like Figure 1 As shown, the shoe upper glue spraying trajectory planning method based on color point cloud data is characterized by comprising the following steps:

[0051] Step S1, obtaining original color point cloud data of a complete shoe upper, and preprocessing the original color point cloud data to obtain color point cloud data of the shoe upper;

[0052] Specifically, a 3D color camera is used to scan the shoe upper at 360 degrees to obtain the original color point cloud data of the shoe upper, and these original color point cloud data are spliced ​​to obtain the following: Figure 2 The original color point cloud data of the complete upper is shown, and the original color point cloud data includes the geometric shape information of the upper and the RGB color information of the upper surface.

[0053] The preprocessing of the original color point cloud data specifically includes:

[0054] Use pass-through filtering to remove Figure 2 The point cloud of the black and white areas in the middle part reduces the point cloud that needs to be processed and obtains the preliminary ROI area containing the shoe upper;

[0055] Since sparse outliers are easily generated during laser scanning, which has a great impact on the subsequent point cloud processing such as the calculation of the normal vector of the sampling point, the statistical filtering method is used to remove outliers in the preliminary ROI area;

[0056] The scanned point cloud is often dense and has a huge number of point clouds. If it is directly calculated and processed, the calculation efficiency is low. Therefore, downsampling is achieved through the voxel grid method, while reducing the number of point clouds, the shape characteristics and color information of the point clouds are maintained, and finally the upper color point cloud data A is obtained, as shown in Figure 3 shown.

[0057] Step S2, dividing the upper color point cloud data into two halves, and rotating one of the halves so that it is located in the same reference plane as the other half, and projecting the point cloud in the reference plane to the same plane to obtain a upper plane image;

[0058] During the laser scanning process, the extracted point cloud color information may be inaccurate or incomplete due to lighting, camera exposure, and other reasons. If the point cloud is directly identified and segmented by the RGB color information of each point in the point cloud, it is difficult to obtain a complete and accurate point cloud area of ​​interest. Since the overall shape of the upper 3D point cloud is a relatively closed multi-surface complex, the outline of the upper is complete and continuous when observed from different angles of the upper, front, back, left, and right, without obvious breaks or missing. If the upper point cloud is directly projected onto a certain reference plane, the projected point cloud can only show the shape and color of the front, and the information on the back of the upper cannot be fully reflected. Therefore, in order to fully display all the surface features and color information of the upper in the projected point cloud, the upper needs to be segmented, and then the segmented point cloud is rotated to the same reference plane by rotation, and then spliced ​​and projected onto the reference plane to obtain complete shoe plane image data.

[0059] The specific operation is as follows: the short axis direction of the upper color point cloud data is the x-axis, the long axis direction is the y-axis, and the vertical upward direction is the z-axis. The upper color point cloud data is divided into two halves along the y-axis direction, and one of the halves is rotated 180° around the rotation axis parallel to the z-axis so that the two halves are located in the same reference plane. The two halves are then projected onto the yz plane to obtain the upper plane image, as shown in Figure 4 shown.

[0060] More specifically, the above operation is implemented by the following steps:

[0061] Step S21, obtaining the maximum value point H (x, y, z) of the toe in the y-axis direction of the upper color point cloud data, taking the plane passing through the maximum value point and parallel to the yz plane as the dividing plane, classifying the point cloud with an x ​​coordinate value less than the x coordinate value of the point H in the upper color point cloud data as point set M, and classifying the point cloud with an x ​​coordinate value greater than the x coordinate value of the point H as point set N, that is, obtaining two halves formed by the point set M and the point set N;

[0062] Step S22, create a rotation axis passing through point H and parallel to the z-axis, rotate the point set N 180° around the rotation axis to obtain the point set O, and merge the point set M and the point set O to obtain the point set Q;

[0063] Step S23: Project the point set Q onto the yz plane, and scale it according to the required resolution and distribution range of the point set Q to obtain a planar image of the upper.

[0064] More specifically, step S23 is implemented by the following steps:

[0065] Step S231, respectively calculating the minimum value min.y and the maximum value max.y of the point set Q in the y dimension, and the minimum value min.z and the maximum value max.z in the z dimension, these values ​​will be used to determine the range of the point cloud in the three-dimensional space, ignoring the x dimension because a two-dimensional color image of the yz plane is created;

[0066] Step S232, creating a two-dimensional color image with a width of col and a height of row, and initializing all pixels of the two-dimensional color image to the background color to distinguish the shoes from the background, wherein the width col and the height row of the two-dimensional color image are calculated according to the formulas row=(max.y-min.y)*scale+2*border and col=(max.z-min.z)*scale+2*border, respectively, scale represents the set image scaling ratio, border represents the set image border size, and scale and border are set according to the required resolution;

[0067] Step S233, calculate the corresponding position (x, y) of each point q in the point set Q on the two-dimensional color image according to the formulas y=border+(qy-min.y)*scale and x=border+(qz-min.z)*scale, and set the RGB value of the corresponding position to the RGB value of point q to obtain the upper plane image, wherein qy represents the y coordinate value of point q, and qz represents the z coordinate value of point q.

[0068] Step S3: Process the upper plane image to obtain a specific area and a center line of the specific area, such as Figure 5 As shown;

[0069] The specific steps include:

[0070] Step S31: To avoid the color extraction effect being affected by factors such as illumination, the RGB color space of the upper plane image is converted to the HSV color space, and an image of the saturation channel in the HSV color space is selected as a processing image;

[0071] Step S32: The processed image is subjected to mean smoothing, grayscale scaling, and binarization in sequence to obtain a high saturation area. In the high saturation area, the hue image is subjected to threshold processing to extract an area of ​​a specific hue, such as Figure 4 The brown stripe area in the image; Among them, mean smoothing can reduce the random noise in the processed image, and gray value scaling can enhance the contrast of saturation information;

[0072] Step S33, performing a filling operation on all regions of a specific hue to form multiple connected domains, selecting the connected domain with the largest area as the target region, and performing closing operations and opening operations on the target region in sequence to fill small holes and remove small protrusions;

[0073] Step S34: extract the contour of the target area processed by step S33, smooth the contour to reduce the jagged effect and noise points on the contour, and generate a filled area according to the smoothed contour to obtain a specific area R, that is, Figure 5 The white area in

[0074] Step S35, applying a Gaussian line extraction algorithm to process the specific area to obtain a line set, and selecting the longest line from the line set as the center line L;

[0075] Step S36: Convert the sub-pixel precision (XLD) contour of the centerline into a point set coordinate set S, and sample the point set coordinate set S at intervals of 30 points to obtain a two-dimensional point set data set T that retains the centerline features. XLD contours are usually used to describe the boundaries of objects in an image, and these boundaries are represented with sub-pixel precision, that is, they can be more accurate than the pixel resolution of the image.

[0076] Step S4: Identify the regional point cloud corresponding to the specific area from the upper color point cloud data, such as Figure 6 As shown;

[0077] The specific steps include:

[0078] Step S41, mapping the points of the point set M and the point set O to the two-dimensional color image respectively, and obtaining the two-dimensional point pixel coordinates corresponding to each point;

[0079] Step S42, counting the points in the point set M whose two-dimensional point pixel coordinates are within the specific area into the point set U, and counting the points in the point set O whose two-dimensional point pixel coordinates are within the specific area into the point set V;

[0080] Step S43, rotating the point set V around the rotation axis in the opposite direction to step S22, rotating 180°, to obtain the point set W;

[0081] Step S44: merge the point set W with the point set U to generate a regional point cloud X corresponding to a specific area, such as Figure 6 shown.

[0082] Step S5, obtaining the trajectory points corresponding to the center line of the specific area in the upper color point cloud data, and calculating the trajectory posture after curve fitting and smoothing the trajectory points to obtain the upper glue spraying trajectory, such as Figure 7 As shown;

[0083] The specific steps include:

[0084] Step S51, according to the formula x=0, y=(ty-border) / scale+min.y and z=(tz-border) / scale+min.z, convert the points in the two-dimensional point set data set T into the three-dimensional point set X, where (x, y, z) represents the coordinates of the coordinate x' of the point in the three-dimensional point set X, ty represents the y coordinate value of the point t in the two-dimensional point set data set T, and tz represents the z coordinate value of the point t in the two-dimensional point set data set T; since the two-dimensional upper plane image is projected onto the yz plane by the three-dimensional point set Q, when converting the points in the two-dimensional point set T into three-dimensional points, their x coordinate values ​​are unknown, so the x coordinate values ​​of all points are initialized to 0 during the conversion;

[0085] Step S52, the x coordinate values ​​of the three-dimensional point set X obtained after the conversion are all 0, which does not fit the shoe point cloud and cannot be directly used as the glue spraying track point. Therefore, it is necessary to calculate the actual x coordinate value of each point. Specifically, for point x' in the three-dimensional point set X, the range between the y coordinate values ​​of the previous point and the next point of point x' is used as the defined range, and the point cloud set within the defined range is searched in the point set O, and the average of the x coordinate values ​​of each point in the point cloud set is calculated as the temporary x coordinate value of point x'; wherein, the y coordinate value of the point cloud in the point cloud set within the defined range is greater than the y coordinate value of point x'-1 and less than the y coordinate value of point x'+1;

[0086] Step S53, obtaining the nearest neighbor point of point x' in the three-dimensional point set X in the point set O, and using the nearest neighbor point as the target point of point x', thereby obtaining a target point set Y corresponding to each point in the three-dimensional point set X;

[0087] Step S54, record the set of points in the target point set Y whose y value is greater than the y value of point H as point set I, record the set of points whose y value is less than the y value of point H as point set J, rotate point set J around the rotation axis in the opposite direction of step S22, rotate 180°, merge the point set obtained after the rotation with point set I, generate a complete trajectory point cloud set, perform B-spline curve fitting and smoothing on the trajectory point cloud set, and obtain an initial glue spraying trajectory point set K;

[0088] Step S55, calculate the normal vector of each point k in the initial glue spraying trajectory point set K (i.e., the local normal direction of the surface where the point is located), and use the normal vector as the X-direction vector (ax, ay, az) of the robot tool. n To point k n+1 The direction is taken as the Y vector (nx, ny, nz) of the robot tool, the Z vector (ox, oy, oz) is calculated using the X vector and the Y vector, and the X vector is corrected using the Y vector and the Z vector to obtain the gluing trajectory of the shoe upper.

[0089] The above description is only a preferred embodiment of the present invention, and therefore cannot be used to limit the scope of implementation of the present invention. That is, equivalent changes and modifications made according to the scope of the patent application of the present invention and the contents of the specification should still fall within the scope covered by the patent of the present invention.

Claims

1. A shoe upper glue spraying trajectory planning method based on color point cloud data, characterized in that: The steps include: Step S1, obtaining original color point cloud data of a complete shoe upper, and preprocessing the original color point cloud data to obtain color point cloud data of the shoe upper; Step S2, dividing the upper color point cloud data into two halves, and rotating one of the halves so that it is located in the same reference plane as the other half, and projecting the point cloud in the reference plane to the same plane to obtain a upper plane image; Step S3, processing the upper plane image to obtain a specific area and a center line of the specific area; Step S4, identifying a regional point cloud corresponding to a specific area from the upper color point cloud data; Step S5, obtaining the trajectory points corresponding to the center line of the specific area in the upper color point cloud data, and calculating the trajectory posture after curve fitting and smoothing the trajectory points to obtain the upper glue spraying trajectory; The step S3 specifically includes the following steps: Step S31, converting the RGB color space of the upper plane image into the HSV color space, and selecting the image of the saturation channel in the HSV color space as the processing image; Step S32, performing mean smoothing, gray value scaling, and binarization on the processed image in sequence to obtain a high saturation area, and performing threshold processing on the hue image in the high saturation area to extract an area with a specific hue; Step S33, performing a filling operation on all regions of a specific hue to form multiple connected domains, selecting the connected domain with the largest area as the target region, and performing closing operations and opening operations on the target region in sequence to fill holes and remove protrusions; Step S34, extracting the contour of the target area processed in step S33, smoothing the contour to reduce the jagged effect and noise points on the contour, and generating a filling area according to the smoothed contour to obtain a specific area; Step S35, applying a Gaussian line extraction algorithm to process the specific area to obtain a line set, and selecting the longest line from the line set as the center line; Step S36: convert the sub-pixel precision contour of the center line into a point set coordinate set S, and sample the point set coordinate set S to obtain a two-dimensional point set data set T that retains the center line features.

2. The method for planning the gluing trajectory of a shoe upper based on color point cloud data according to claim 1, characterized in that: In step S1, a 3D color camera is used to perform a 360-degree scan on the shoe upper to obtain original color point cloud data of the complete shoe upper, where the original color point cloud data includes geometric shape information of the shoe upper and RGB color information of the shoe upper surface.

3. The method for planning the gluing trajectory of a shoe upper based on color point cloud data according to claim 1, characterized in that: In the step S1, the preprocessing of the original color point cloud data specifically includes: using a straight-through filtering method to obtain a preliminary ROI area containing the upper, using a statistical filtering method to remove outliers in the preliminary ROI area and implementing downsampling through a voxelized grid method to obtain the upper color point cloud data A.

4. A method for planning the gluing trajectory of a shoe upper based on color point cloud data according to claim 1, 2 or 3, characterized in that: In step S2, the short axis direction of the upper color point cloud data is the x-axis, the long axis direction is the y-axis, and the vertical upward direction is the z-axis. The upper color point cloud data is divided into two halves along the y-axis direction, and one of the halves is rotated 180° around a rotation axis parallel to the z-axis so that the two halves are located in the same reference plane, and then the two halves are projected onto the yz plane to obtain the upper plane image.

5. The method for planning the gluing trajectory of a shoe upper based on color point cloud data according to claim 4, characterized in that: The step S2 specifically includes the following steps: Step S21, obtaining the maximum value point H of the toe in the y-axis direction of the upper color point cloud data, taking the plane passing through the maximum value point and parallel to the yz plane as the dividing plane, classifying the point cloud with an x ​​coordinate value less than the x coordinate value of the point H in the upper color point cloud data as the point set M, and classifying the point cloud with an x ​​coordinate value greater than the x coordinate value of the point H as the point set N, that is, obtaining two halves formed by the point set M and the point set N; Step S22, create a rotation axis passing through point H and parallel to the z-axis, rotate the point set N 180° around the rotation axis to obtain the point set O, and merge the point set M and the point set O to obtain the point set Q; Step S23: Project the point set Q onto the yz plane, and scale it according to the required resolution and distribution range of the point set Q to obtain the upper plane image.

6. The method for planning the gluing trajectory of a shoe upper based on color point cloud data according to claim 5, characterized in that: The step S23 specifically includes the following steps: Step S231, respectively calculating the minimum value min.y and the maximum value max.y of the point set Q on the y-axis, and the minimum value min.z and the maximum value max.z on the z-axis; Step S232, create a two-dimensional color image with a width of col and a height of row, and initialize all pixels of the two-dimensional color image to the background color, wherein the width col and the height row of the two-dimensional color image are calculated according to the formulas row=(max.y-min.y)*scale+2*border and col=(max.z-min.z)*scale+2*border respectively, scale represents the set image scaling ratio, border represents the set image border size, and scale and border are set according to the required resolution; Step S233, calculate the corresponding position (x1, y1) of each point q in the point set Q on the two-dimensional color image according to the formulas y1=border+(qy-min.y)*scale and x1=border+(qz-min.z)*scale, and set the RGB value of the corresponding position to the RGB value of point q to obtain the upper plane image, wherein qy represents the y coordinate value of point q, and qz represents the z coordinate value of point q.

7. The method for planning the gluing trajectory of a shoe upper based on color point cloud data according to claim 5, characterized in that: The step S4 specifically includes the following steps: Step S41, mapping the points of the point set M and the point set O to the two-dimensional color image respectively, and obtaining the two-dimensional point pixel coordinates corresponding to each point; Step S42, counting the points in the point set M whose two-dimensional point pixel coordinates are within the specific area into the point set U, and counting the points in the point set O whose two-dimensional point pixel coordinates are within the specific area into the point set V; Step S43, rotating the point set V around the rotation axis in the opposite direction to step S22, rotating 180°, to obtain the point set W; Step S44: Merge the point set W with the point set U to generate a regional point cloud X corresponding to the specific area.

8. The method for planning the gluing trajectory of a shoe upper based on color point cloud data according to claim 6, characterized in that: The step S5 specifically includes the following steps: Step S51, according to the formulas x2=0, y2=(ty-border) / scale+min.y and z2=(tz-border) / scale+min.z, convert the points in the two-dimensional point set data set T into the three-dimensional point set X, where (x2, y2, z2) represents the coordinates of the point x' in the three-dimensional point set X, ty represents the y coordinate value of the point t in the two-dimensional point set data set T, and tz represents the z coordinate value of the point t in the two-dimensional point set data set T; Step S52: for a point x' in the three-dimensional point set X, the range between the y coordinate values ​​of the previous point and the next point of the point x' is used as a bounding range, a point cloud set within the bounding range is searched in the point set O, and the average of the x coordinate values ​​of each point in the point cloud set is calculated as a temporary x coordinate value of the point x'; Step S53, obtaining the nearest neighbor point of point x' in the three-dimensional point set X in the point set O, and using the nearest neighbor point as the target point of point x', thereby obtaining a target point set Y corresponding to each point in the three-dimensional point set X; Step S54, record the set of points in the target point set Y whose y value is greater than the y value of point H as point set I, record the set of points whose y value is less than the y value of point H as point set J, rotate point set J around the rotation axis in the opposite direction of step S22, rotate 180°, merge the point set obtained after the rotation with point set I, generate a complete trajectory point cloud set, perform B-spline curve fitting and smoothing on the trajectory point cloud set, and obtain an initial glue spraying trajectory point set K; Step S55: Calculate the normal vector of each point k in the initial glue spraying trajectory point set K. The normal vector is used as the X-direction vector of the robot tool. n To point k n+1 The direction of the robot tool is taken as the Y direction vector, the Z direction vector is obtained by using the X direction vector and the Y direction vector, and the X direction vector is corrected by using the Y direction vector and the Z direction vector to obtain the gluing trajectory of the shoe upper.

Images