Intelligent vamp side edge recognition and line drawing method and device and storable medium

Abstract

The invention provides an intelligent vamp side edge recognition and line drawing device and method.The intelligent vamp side edge recognition and line drawing device comprises a high-precision linear module, a laser scanner, a V-shaped stop block, a pneumatic gripper, a control and data acquisition unit and an industrial robot which are all arranged on a vibration-resistant optical platform; the high-precision linear module is used for controlling a shoe body to move at a constant speed, the laser scanners are arranged on the two sides of the high-precision linear module, the pneumatic gripper and the V-shaped baffle are installed above the high-precision linear module, and each laser scanner comprises a line laser camera and a line laser base. And the control and data acquisition unit is connected with the vibration-resistant optical platform, controls the shoe body to move at a constant speed, performs line laser scanning, receives the point cloud number, processes the point cloud data by using a self-adaptive growth algorithm and a fracture filling algorithm to obtain a vamp side closed curve, and controls the industrial robot to trace lines on the vamp side. The shoe body line drawing precision and efficiency are improved, the shoe body line drawing device can be suitable for curved surfaces and lines of various shoe types, the requirement for mass production is met, and the line drawing path continuity and smoothness are guaranteed.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of shoemaking equipment, in particular to an intelligent vamp side edge recognition and line drawing method, device and storage medium. BACKGROUND

[0002] In the vamp production process, the traditional line drawing (such as Figure 1 indicated) that accurately draws color or pattern lines on the vamp relies on manual or semi-automatic equipment, and has the problems of limited precision, low efficiency and poor repeatability. The existing vamp line drawing method includes a line drawing machine method, a preset processing trajectory method, and a structured light scanning and curved edge detection method.

[0003] As shown in Figure 2 , the line drawing machine method is to press different vamp patterns on the transmission platform, use camera vision positioning, and draw according to the preset pattern. However, the line drawing machine belongs to pre-sewing equipment and can only draw lines on the pattern and position in a two-dimensional plane, and cannot be applied to the line drawing process after molding.

[0004] The preset processing trajectory method is to measure the size data of the processing area of different shoe models, generate different preset trajectories for matching, and then use a mechanical arm or other equipment to draw lines along the preset trajectory. However, this method cannot automatically adapt to errors generated in flexible processing, the positioning accuracy is difficult to guarantee, and it relies on manual operation or semi-automatic tools, so it is difficult to achieve both precision and efficiency.

[0005] The structured light scanning and curved edge detection method is to project structured light onto the vamp, collect in real time by a binocular camera or a special three-dimensional camera, generate high-density point clouds, extract the curved profile from the point clouds by using two-dimensional projection techniques such as Canny edge detection and normal vector change, and map the profile back to the three-dimensional space to generate the path. However, this method cannot identify non-edge areas, is sensitive to high-frequency noise, cannot automatically compensate for cracks or burrs, and ultimately results in gaps or breakpoints in the path, so the efficiency of processing a large number of similar shoe models is low.

[0006] In industrial automatic line drawing applications, the vamp surface curvature changes complexly, the line shape is diverse, and the reflection or texture of different vamp materials interferes, making it difficult for simple threshold segmentation or two-dimensional projection-based methods to meet the needs of high robustness and high precision, and lacking local smoothing. In addition, in the case of noise and scanning occlusion, the broken or discontinuous curve is automatically completed to ensure closure.

[0007] There is a large room for improvement in how to obtain high-quality point clouds, accurately extract line regions, implement crack completion and closure fitting, and generate high-precision robot paths in the prior art. Therefore, based on the above problems, the present application proposes an intelligent vamp side edge recognition and line drawing method, device and storage medium. SUMMARY

[0008] The present application is directed to one or more technical defects in the prior art, and proposes the following technical solutions.

[0009] Based on the first aspect of the present application, an intelligent shoe side edge recognition and line drawing device is proposed, which comprises a high-precision linear module, a laser scanner, a V-shaped block, a pneumatic gripper, a control and data acquisition unit, and an industrial robot.

[0010] The high-precision linear module, laser scanner, V-shaped block, pneumatic gripper and industrial robot are all deployed on a vibration-resistant optical platform.

[0011] The high-precision linear module is used to control the uniform motion of the shoe body, and the laser scanner is deployed on both sides of the high-precision linear module. The pneumatic gripper and the V-shaped block are installed above the high-precision linear module. The laser scanner comprises a line laser camera and a line laser base, and the line laser camera is installed on the line laser base for emitting high-density line laser and acquiring laser profile data.

[0012] The control and data acquisition unit is connected with the vibration-resistant optical platform, and comprises an embedded controller and a data processing host, which are used to synchronously control the uniform motion of the shoe body, line laser scanning, receive point cloud data, and

[0013] The point cloud data is processed by using adaptive growth algorithm and fracture filling algorithm to obtain a closed curve of the shoe side edge, and the industrial robot is controlled according to the closed curve of the shoe side edge to complete the shoe side edge line drawing.

[0014] The high-precision linear module of the device ensures the uniform motion of the shoe body, provides a stable basis for laser scanning, avoids point cloud distortion caused by motion jitter, and enables the bilateral laser scanners to work synchronously to eliminate the blind area caused by unilateral shielding and collect complete point cloud data in all directions. The shoe body is flexibly fixed to prevent sliding and ensure consistency during scanning and line drawing.

[0015] Furthermore, the pneumatic gripper and the V-shaped block are respectively located on the front and rear sides of the shoe body to fix the shoe body on the high-precision linear module for uniform motion.

[0016] When the high-precision linear module moves at a constant step, the laser scanner emits high-density line laser on the surface of the shoe body and acquires corresponding three-dimensional sampling points of the shoe body to collect comprehensive high-density point cloud of the shoe body.

[0017] This step adopts anti-interference design to isolate environmental vibration interference and ensure uniform point cloud density and cover the entire area of the shoe upper.

[0018] Based on the second aspect of the present application, a method for shoe side edge recognition and delineation using the device described in any of the above is also proposed, comprising:

[0019] S1: Collecting point cloud data of a shoe body by the intelligent shoe side edge recognition and delineation device, performing scalar field conversion, pre-processing the point cloud data to obtain a plurality of connected sub-regions and a candidate point set of a region to be delineated;

[0020] S2: Expanding the region to be delineated candidate point set by an adaptive growing algorithm. Specifically, based on a preset radius, a seed point is found in the KD tree constructed from the region to be delineated candidate point set. The score of each point in the KD tree is calculated from the seed point, and the two neighbor points with the highest scores are selected to join the next round of growing queue. The growing direction is updated. The longest connected domain is selected as the optimal backbone point by the main direction projection cumulative length and the noise is removed to obtain a continuous backbone point set.

[0021] S3: Completing the breakpoints in the backbone point set by a breakage filling algorithm. Specifically, the distance from the midpoint of the two adjacent breakpoints to the centroid is calculated, the curvature is selected in combination with the average radius of the contour, the Bezier peak point is set in combination with the curvature, the insertion point sequence between the two breakpoints is generated by quadratic Bezier interpolation, the normal vector of the insertion point sequence is smoothly transitioned by spherical linear interpolation, and the insertion point sequence is processed by Gaussian weighted smoothing to obtain a shoe side edge closed curve.

[0022] S4: The industrial robot performs shoe side edge delineation by taking the shoe side edge closed curve as the processing path.

[0023] Further, the pre-processing of the point cloud data includes assigning attribute values to each point, setting a preliminary threshold filtering condition according to multi-dimensional attribute values, removing noise points, and extracting connected domains based on octree or K-nearest neighbor principle.

[0024] This step can realize accurate extraction of the region to be delineated in high-density point cloud, ensure effective removal of non-target regions, improve the recognition rate and accuracy of the region to be delineated, and reduce invalid calculations.

[0025] Further, after step S1 and before step S2, the point cloud data discrete set is divided into a plurality of connected sub-regions, and the point cloud main direction is calculated. It is evaluated whether the morphology of each connected sub-region meets the delineable feature. The point cloud data in the connected sub-region that meets the delineable feature is taken as the candidate point set to be delineated, and the point cloud data that does not meet the delineable feature is filtered.

[0026] Further, the score calculation formula of each point is:

[0027]

[0028] wherein, denotes the last growth direction of the current point, denotes the direction vector from the current point to the neighbor point, p neigh denotes the candidate neighbor point, p curr denotes the current growth point, d = ||p neigh -p curr ||, d denotes the Euclidean distance.

[0029] Further, the formula for updating the growth direction is:

[0030]

[0031] wherein, denotes the last growth direction of the current point, denotes the direction vector from the current point to the neighbor point, denotes the new growth direction of the current point;

[0032] The formula for calculating the main direction projection cumulative length is:

[0033]

[0034] wherein, M denotes the total number of points in the connected domain, Pt denotes the tth three-dimensional point coordinate in the connected domain according to the main direction sorting, and Pt+1 denotes the next point adjacent to Pt after sorting according to the main direction.

[0035] This step can preferentially select neighbors with similar directions and close distances, effectively avoid path bifurcation, ensure direction consistency, make the growth trajectory naturally fit the vamp curvature, inhibit the jump of normal vector, and ensure that the effective curve is retained.

[0036] Further, the points in the connected sub-region are projected onto the main direction line, the polar angle is sorted and outliers are removed, and if the distance between adjacent polar angle sorted points is greater than a preset breaking threshold, a break exists at this place.

[0037] This step can improve the accuracy of point cloud detection and adapt to different shoe type point cloud densities.

[0038] Further, when the contour average radius is less than the distance from the midpoint of the adjacent two breaking points to the centroid, the forefoot curvature is selected as the curvature, otherwise, when the contour average radius is greater than the distance from the midpoint of the adjacent two breaking points to the centroid, the heel curvature is selected as the curvature.

[0039] The step improves the fitting degree of the interpolation sequence by curvature self-adaptation, completes sharp arc lines by foot head curvature, and completes smooth arc lines by foot heel curvature, and can ensure the continuity and smoothness of the closed curve and the drawing path after the supplement of the points.

[0040] Further, the formula for performing the quadratic Bezier interpolation is:

[0041] B(t) = (1-t) 2 s+2(1-t)tp ctrl +t 2 e,t∈[0,1]

[0042] The calculation formula for setting the Bezier peak point in combination with the curvature is:

[0043] p ctrl = m-(κL)u

[0044] The calculation formula for the Gaussian weighting is:

[0045]

[0046] wherein p ctrl represents the peak point, κ represents the curvature, m represents, L represents the distance between the two breaking points, s and e represent the breaking start point and the breaking end point, t represents the interpolation parameter, represents the coordinate of the interpolation point sequence after the smoothing processing, B(t) represents the coordinate of the interpolation point sequence, M represents the total length of the interpolation point sequence, k represents the index of the resampling point, and σ represents the standard deviation of the Gaussian function.

[0047] Based on the third aspect of the application, a computer program product is also provided, which has one or more computer programs thereon, and when the computer programs are executed by a computer processor, the method according to any one of the above aspects is implemented.

[0048] The technical effect of the application is that the application adopts bilateral line laser scanning in the motion state, realizes panoramic point cloud data acquisition, can ensure data integrity and consistency, proposes noise points and non-target regions based on an adaptive growth algorithm, improves the recognition rate and precision of the region to be ruled, and guarantees the continuity and smoothness of the processing path through the breaking filling algorithm and the closed curve fitting smoothing technology. The application has high integration, can realize automatic closed loop operation, and well meets the large batch production and diversified, high precision intelligent drawing requirements of footwear. BRIEF DESCRIPTION OF DRAWINGS

[0049] Other features, objects and advantages of the application will become more apparent from the following detailed description of non-limiting embodiments made with reference to the accompanying drawings.

[0050] Figure 1is a hand-drawn line drawing provided according to the prior art.

[0051] Figure 2 is a shoe upper line drawing machine drawing provided according to the prior art.

[0052] Figure 3 is a structure diagram of an intelligent shoe upper side edge recognition and line drawing device provided according to an embodiment of the present application.

[0053] Figure 4 is a calibration flowchart of an intelligent shoe upper side edge recognition and line drawing device provided according to an embodiment of the present application.

[0054] Figure 5 is a flowchart of an intelligent shoe upper side edge recognition and line drawing method provided according to an embodiment of the present application.

[0055] Figure 6 is a point cloud processing flowchart of an intelligent shoe upper side edge recognition and line drawing method provided according to an embodiment of the present application.

[0056] Figure 7 is a structure diagram of a computer system of an electronic device suitable for implementing an embodiment of the present application. DETAILED DESCRIPTION

[0057] The present application will be further described in detail below with reference to the accompanying drawings and embodiments. It can be understood that the specific embodiments described herein are only used to explain the related application, and are not a limitation on the application. In addition, it should be noted that, for the sake of description, only the parts related to the application are shown in the drawings.

[0058] It should be noted that the embodiments in the present application and the features in the embodiments can be combined with each other without conflict. The present application will be described in detail below with reference to the accompanying drawings and embodiments.

[0059] Reference Figure 3 which shows an intelligent shoe upper side edge recognition and line drawing device, comprising a high-precision straight line module 1, a laser scanner, a V-shaped block 2, a pneumatic grab 4, a control and data acquisition unit (not shown in the figure) and an industrial robot 8;

[0060] The high-precision straight line module 1, the laser scanner, the V-shaped block 2, the pneumatic grab 4, the control and data acquisition unit and the industrial robot 8 are all deployed on the vibration-proof optical platform 7;

[0061] The high-precision linear module 1 is used for controlling the uniform motion of the shoe body, the laser scanner is arranged on both sides of the high-precision linear module 1, the pneumatic gripper 4 and the V-shaped baffle 2 are installed above the high-precision linear module 1, the laser scanner comprises a line laser camera 3 and a line laser base 6, the line laser camera 3 is installed on the line laser base 6 and is used for emitting high-density line laser and acquiring laser profile data;

[0062] The control and data acquisition unit is connected with the vibration-resistant optical platform and comprises an embedded controller and a data processing host, is used for synchronously controlling the uniform motion of the shoe body, line laser scanning, receiving point cloud data, and the like,

[0063] The point cloud data is processed by using an adaptive growth algorithm and a fracture filling algorithm, a vamp side edge closed curve is obtained, and the industrial robot 8 is controlled according to the vamp side edge closed curve to complete vamp side edge line drawing.

[0064] It should be noted that the pneumatic gripper 4 and the V-shaped baffle 2 are respectively located on the front and rear sides of the shoe body 5, and the shoe body 5 is fixed on the high-precision linear module 1 to move at a uniform speed.

[0065] When the high-precision linear module 1 moves at a constant step and a uniform speed, the laser scanner emits high-density line laser on the surface of the shoe body 5 and acquires corresponding three-dimensional sampling points of the shoe body, and comprehensive high-density point cloud of the shoe body is collected.

[0066] It should be noted that the repeated positioning accuracy of the high-precision linear module 1 is within 5 μm, which is used for stably controlling the accurate motion of the shoe body in the horizontal direction and realizing stable conveying in the scanning process.

[0067] The repeated accuracy of the bilateral line laser scanner is 1 μm, the measurement range is 245±34 mm, the width is 72 mm, the bilateral line laser scanner is respectively arranged on the left and right sides of the module, emits high-density line laser when moving linearly along the vamp, acquires laser profile data of two visual angles, the bilateral arrangement can eliminate unilateral shielding, reduce scanning process time and spatial complexity, and realize fast global uniform scanning.

[0068] The pneumatic gripper fixes the shoe body and can prevent the shoe body from sliding during the motion of the linear module.

[0069] The control and data acquisition unit comprises a high-performance embedded controller and a data processing host, is used for synchronously controlling the motion of the linear module, line laser scanning and point cloud data receiving.

[0070] It should be noted that when the linear module moves at a constant step at a constant speed, the bilateral line laser scanner continuously projects line light on the surface of the vamp, and the line laser sensor automatically acquires corresponding three-dimensional sampling points inside. The left and right line laser sensors synchronize the time stamp of the continuous motion steps, and real-time acquisition of a series of point cloud frame sets is realized, so as to realize comprehensive high-density point cloud collection of the left and right shoes.

[0071] In a specific embodiment, point cloud data is collected simultaneously at three different positions using a ball with a diameter of 50 mm, the position transformation matrix of the ball center in the two line lasers is obtained, and the average value is taken, and the transformation matrix is registered and verified using a standard ball with a smaller diameter, and the spatial relationship of the double line lasers is determined by ball center fitting.

[0072] It should be noted that the application uses high-precision linear modules and bilateral line lasers to simultaneously scan the left and right vamps, and can obtain complete and balanced three-dimensional point cloud data. The motion platform, laser scanning, point cloud processing and robot line drawing module are integrated, the line laser, linear module and robot nozzle are calibrated with high precision, the closed-loop integration from three-dimensional data to automatic line drawing is realized, and the high-precision point cloud model is constructed by combining subsequent line laser calibration, downsampling and normal vector calculation and other preprocessing algorithms.

[0073] It should be noted that the process of robot end calibration is as shown in Figure 4 The line laser calibration is as described above, the robot and the high-precision linear module use the calibration method of eye outside hand, and the mechanical arm end coordinate system is represented as O b -X b Y b Z b , the measurement coordinate system of the line laser is represented as O c -X c Y c Z c , the mechanical arm end coordinate system is represented as O t -X t Y t Z t , the corner point on the linear module is selected, represented as P c in the line laser coordinate system, the mechanical arm end probe touches the corner point, and the coordinates of the corner point in the mechanical arm base coordinate system P b are recorded.

[0074] The transformation matrix of the line laser coordinate system to the mechanical arm end probe coordinate system is The transformation matrix of the mechanical arm end probe coordinate system to the mechanical arm base coordinate system can be obtained from the mechanical arm end pose parameters, and is represented as Then: The above formula is extended to the homogeneous form:

[0075]

[0076] Let and set Then:

[0077]

[0078] For each pair of corresponding points (P c ,P b ), three linear equations are obtained, and the unknown vector is set as Integrating n pairs of feature points can be written as AX = B, and the least squares method is used to solve: X = (A T A) -1 A T B;

[0079] Select three different positions to touch three times at different angles, and the average result can complete the calibration process of the robot end.

[0080] The following refers to Figure 5 , Figure 5 A method for identifying and drawing the side edge of a vamp using the device described in any of the above is shown, comprising:

[0081] S1: Collecting point cloud data of a shoe body by the intelligent vamp side edge identification and drawing device, performing scalar field conversion, and pre-processing the point cloud data to obtain a plurality of connected sub-regions and a candidate point set of a region to be drawn;

[0082] S2: Expanding the region to be drawn by an adaptive growing algorithm, specifically, based on a preset radius, finding a seed point in the KD tree constructed from the candidate point set of the region to be drawn, calculating the score of each point in the KD tree from the seed point, and selecting the two neighbor points with the highest scores to join the next round of growing queue, updating the growing direction, and selecting the longest connected domain as the optimal backbone point through the main direction projection cumulative length and eliminating noise to obtain a continuous backbone point set;

[0083] S3: Completing the breakpoints in the backbone point set by a broken filling algorithm, specifically, calculating the distance from the midpoint of two adjacent broken points to the centroid, selecting curvature in combination with the average radius of the contour, setting Bezier peak points in combination with the curvature, generating an insertion point sequence between the two broken points through quadratic Bezier interpolation, smoothing the normal vector of the insertion point sequence through spherical linear interpolation, and performing Gaussian weighted smoothing processing on the insertion point sequence to obtain a closed curve of the vamp side edge;

[0084] S4: The industrial robot draws the vamp side edge as a processing path based on the closed curve of the vamp side edge.

[0085] It should be noted that the formula for performing quadratic Bezier interpolation is:

[0086] B(t) = (1-t) 2 s+2(1-t)tp ctrl +t 2 e,t∈[0,1]

[0087] The calculation formula for setting the Bezier peak point in combination with the curvature is:

[0088] p ctrl = m-(κL)u

[0089] The calculation formula of the Gaussian weighting is:

[0090]

[0091] Where p ctrl represents the peak point, κ represents the curvature, m represents, L represents the distance between the two fracture points, s, e represents the fracture starting point and the fracture end point, t represents the interpolation parameter, represents the smoothed interpolation point sequence coordinates, B(t) represents the interpolation point sequence coordinates, M represents the total length of the interpolation point sequence, k represents the index of the resampling point, and σ represents the standard deviation of the Gaussian function.

[0092] It should be noted that, as Figure 6 shown, the preprocessing of the point cloud data includes assigning attribute values to each point, such as the height from the shoe upper reference plane, the normal vector and the curvature, and the feature calculation value, setting a preliminary threshold filtering condition according to the multi-dimensional attribute value, such as removing noise points with curvature >0.5 or height exceeding ±5mm, obtaining a rough discrete point set (380,000 points) containing candidate points for the marking area, extracting connected domains based on octree or K-nearest neighbor principle, dividing the discrete point set into multiple connected sub-regions, retaining 12 sub-regions with point number >500, calculating the main direction of the point cloud for each connected sub-region, filtering the point cloud data that does not meet the adaptive threshold using the growth algorithm, and performing fine extraction on the point cloud data that meets the adaptive threshold evaluation, supplementing and smoothing the fracture, obtaining a closed curve and then converting it into a machining path of the robot.

[0093] It should be noted that after step S1 and before step S2, the point cloud data discrete set is divided into multiple connected sub-regions, and the main direction of the point cloud is calculated, and whether the shape of each connected sub-region meets the markable feature is evaluated, the point cloud data in the connected sub-region that meets the markable feature is taken as a candidate point set for marking, and the point cloud data that does not meet the markable feature is filtered.

[0094] It should be noted that the score calculation formula of each point is:

[0095]

[0096] wherein, denotes the last growth direction of the current point, denotes the direction vector from the current point to the neighbor point, p neigh denotes the candidate neighbor point, p curr denotes the current growth point, d = ||p neigh -p curr ||, d denotes the Euclidean distance.

[0097] It should be noted that the formula for updating the growth direction is:

[0098]

[0099] wherein, denotes the last growth direction of the current point, denotes the direction vector from the current point to the neighbor point, denotes the new growth direction of the current point;

[0100] The formula for calculating the main direction projection cumulative length is:

[0101]

[0102] wherein, M denotes the total number of points in the connected domain, Pt denotes the tth three-dimensional point coordinate in the connected domain sorted according to the main direction, and Pt+1 denotes the next point adjacent to Pt after sorting according to the main direction.

[0103] It should be noted that the points in the connected sub-region are projected onto the main direction line, polar angle sorting is performed and outliers are removed, and if the distance between adjacent polar angle sorting points is greater than a preset breaking threshold, a break exists at this place.

[0104] It should be noted that when the contour average radius is less than the distance from the midpoint of the adjacent two breaking points to the centroid, the toe curvature is selected as the curvature, otherwise, when the contour average radius is greater than the distance from the midpoint of the adjacent two breaking points to the centroid, the heel curvature is selected as the curvature.

[0105] In specific embodiments, KD trees are constructed for different radii (r = 1, r = 1.5, r = 2) in sequence, a seed point A (point with maximum curvature) with degree 1 is found within the radius range, the score of each point in the KD tree is calculated according to the score formula from the seed point A, the neighbor points with an included angle less than 40° are reserved in combination with the surface straight line or curve characteristics of the shoe body, the top two neighbor points with the highest scores are added to the next round queue, and the growth direction is updated;

[0106] From the seed point, gradually expand and absorb 8000 points of B, C, D, etc. More than 7500, the growth result point number is too much, indicating that the current growth is over-expanding, then enumerate different connection degrees (neighbor number 6-10) to extract connected domains, filter the noise points of loose connection, calculate the main direction projection cumulative length of each connected domain, and select the connected domain with the longest main direction projection length as the optimal trunk;

[0107] Within the radius range of r=1.5, the number of trunk neighbors and non-trunk neighbors is counted, and if the non-trunk neighbor number of a certain point is too high and the average distance is small, or the branch angle is > 25°, the point is removed, and the graph region is obtained;

[0108] Further segmentation of the graph region is performed, the local region of the point cloud is projected onto the orthogonal plane of the main direction to obtain an approximately two-dimensional projection graph, and the graph center or the estimated starting point is taken as the center to expand along both sides of the main direction. When expanding, it is determined whether to include the same sub-region according to the distance between the projection point and the segmentation straight line or curve, and the two-dimensional segmentation result is mapped to the three-dimensional space to obtain a more accurate target region point cloud.

[0109] In a specific embodiment, after the point cloud is converted into a two-dimensional main coordinate system, the mean value and standard deviation of the radial value are calculated, and outliers with an absolute value greater than a threshold value between the mean value and the standard deviation are removed;

[0110] The distance d of adjacent ordered points is calculated i and the average distance The fracture threshold is set If the distance of multiple adjacent ordered points is greater than the fracture threshold, then the index i→i+1 has a fracture;

[0111] For each fracture endpoint pair (s, e), the distance p of the midpoint m=(s+e) / 2 to the centroid is calculated, and compared with the average radius of the contour If Select the "toe curvature" κ toe , otherwise select the "heel curvature" κ heel , calculate the endpoint distance L=‖e-s‖, the unit vector u of the midpoint of the fracture endpoint to the centroid direction, and after plane projection, perform normalization processing, and then perform smoothing processing. Specifically, set the Bezier peak point p ctrl =m-(κL)u, and according to the quadratic Bezier formula,

[0112] B(t)=(1-t) 2 s+2(1-t)tp ctrl +t 2 e,t∈[0,1], insert several points of t=k / (K-1)(k=0,…,K-1), and the normal vector is sequentially interpolated in n s and ne Smooth transition between, the interpolated ordered point sequence {q i}, under the cycle index, with a window size of 7 Gaussian weighting:

[0113]

[0114] The corresponding normal vector is also weighted and normalized, and multiple iterations are performed to obtain a smoother curve. The cumulative arc length sequence {s i} of the smoothed closed curve is calculated, the total arc length s M , the target point number N t , and the arc length equal position Linear interpolation on {s i} to obtain resampled points q ' k The normal vector can be directly copied if the included angle at both ends is close to 0°, otherwise spherical interpolation is also performed to obtain a closed curve.

[0115] It should be noted that in the process of robot processing, in order to make the processing path conform to the original area information to the greatest extent, the line slice segmentation method is used to find the centroid. Specifically, equidistant slicing is performed along the closed curve at a set step length, the cross section is obtained, and the point cloud or fitted curve of each section is projected in two dimensions. The geometric centroid of the section point set is calculated and reflected in the three-dimensional space, that is, the path point of the section. The corresponding pose of each path point is calculated using the point cloud normal vector information, so that the robot can maintain the constant verticality or preset inclination between the nozzle and the upper during processing.

[0116] It should be noted that the transformation matrix of the three-dimensional path point and the pose through the line laser coordinate system to the robot base coordinate system, and the target trajectory point of the robot vision coordinate system, can make the robot accurately outline the shoe body.

[0117] It should be noted that the present application adopts bilateral line laser scanning in a motion state to realize panoramic point cloud data acquisition, which can ensure data integrity and consistency. Based on the adaptive growth algorithm, noise points and non-target regions are proposed, which improves the recognition rate and accuracy of the to-be-ruling region. Through the fracture filling algorithm and closed curve fitting smoothing technology, the continuity and smoothness of the processing path are ensured. The present application has high integration and can realize automatic closed-loop operation, which well meets the mass production and diversified, high-precision intelligent ruling requirements of footwear.

[0118] It should be noted that the high-precision straight line module in the device of the present application ensures that the shoe body maintains uniform motion, providing a stable basis for laser scanning and avoiding motion jitter leading to point cloud loss.

[0119] The bilateral laser scanners work synchronously, can eliminate the single-sided blind area, collect complete point cloud data in all directions, and prevent the shoe body from sliding by flexibly fixing the shoe body, so that the shoe body remains consistent during scanning and line drawing, and the anti-interference design can isolate the interference of environmental vibration, ensure uniform point cloud density, and cover all areas of the vamp.

[0120] It should be noted that the application can realize accurate extraction of the to-be-lining region in the high-density point cloud, ensure effective elimination of non-target regions, improve the recognition rate and accuracy of the to-be-lining region, reduce invalid calculation, preferentially select neighbors with similar directions and close distances, effectively avoid path bifurcation, ensure direction consistency, make the growth trajectory naturally fit the vamp curvature, inhibit normal vector jumping, ensure effective curves are retained, improve the accuracy of point cloud detection, and adapt to point cloud densities of different shoe types.

[0121] Reference will be made below to the accompanying drawings Figure 7 which shows a structural schematic diagram of a computer system 700 of an electronic device suitable for implementing embodiments of the application. Figure 7 The electronic device shown is merely an example and should not impose any limitation on the functions and use range of embodiments of the application.

[0122] As shown in Figure 7 , the computer system 700 includes a central processing unit (CPU) 701 which can perform various appropriate actions and processes according to programs stored in a read-only memory (ROM) 702 or programs loaded from a storage portion 708 into a random access memory (RAM) 703. In the RAM 703, various programs and data required for the operation of the system 700 are also stored. The CPU 701, the ROM 702, and the RAM 703 are connected to each other through a bus 704. An input / output (I / O) interface 705 is also connected to the bus 704.

[0123] The following components are connected to the I / O interface 705: an input portion 706 including a keyboard, a mouse, and the like; an output portion 707 including a liquid crystal display (LCD), a speaker, and the like; a storage portion 708 including a hard disk, and the like; and a communication portion 709 including a network interface card such as a LAN card, a modem, and the like. The communication portion 709 performs communication processing via a network such as the Internet. A drive 710 is also connected to the I / O interface 705 as needed. A removable medium 711 such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, and the like is mounted on the drive 710 as needed, so that a computer program read therefrom is installed in the storage portion 708 as needed.

[0124] In particular, the processes described above with reference to the flow charts can be implemented as a computer software program according to embodiments of the present disclosure. For example, embodiments of the present disclosure include a computer program product comprising a computer program carried on a computer readable storage medium, the computer program comprising program code for performing the methods illustrated by the flow charts. In such embodiments, the computer program can be downloaded and installed from a network via the communication section 709, and / or installed from the removable media 711. When the computer program is executed by the central processing unit (CPU) 701, the above-described functions defined in the methods of the present application are performed. It should be noted that the computer readable storage medium of the present application can be a computer readable signal medium or a computer readable storage medium or any combination of the two. The computer readable storage medium may, for example, be - but is not limited to - an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus or device, or any suitable combination of the above. More specific examples of the computer readable storage medium can include, but are not limited to, an electrical connection having one or more wires, a portable computer diskette, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present application, the computer readable storage medium can be any tangible medium that contains or stores a program that can be used by or in connection with an instruction execution system, apparatus or device. In the present application, the computer readable signal medium can include a data signal carried in a baseband or as part of a carrier wave, in which the computer readable program code is carried. Such a propagated data signal can take any of a variety of forms, including but not limited to electro-magnetic, optical, or any suitable combination thereof. The computer readable signal medium can also be any computer readable storage medium that can be used for by or in connection with an instruction execution system, apparatus or device, and that can contain or store a program for use by or in connection with an instruction execution system, apparatus or device. The program code contained on the computer readable storage medium can be transmitted by any suitable medium, including but not limited to wireless, wire line, optical fiber cable, RF, etc., or any suitable combination of the above.

[0125] Computer program code for carrying out operations of the present application can be written in any combination of one or more programming languages, including an object oriented programming language such as Java, Smalltalk, C++ or the like, and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code can execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or server. In the latter scenario, the remote computer can be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or the connection can be made to an external computer (for example, through the Internet using an Internet Service Provider).

[0126] The computer program instructions can also be loaded onto a computer, other programmable data processing apparatus, or other devices to cause a series of operational steps to be performed on the computer, other programmable apparatus or other devices to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide processes for implementing the functions / acts specified in the flowchart and / or block diagram block or blocks.

[0127] The modules involved in the embodiments of the present application can be implemented in the form of software or hardware.

[0128] As another aspect, the application also provides a computer readable storage medium, which can be included in the electronic device described in the above embodiments, or can exist independently without being assembled into the electronic device. The computer readable storage medium carries one or more programs, which, when executed by the electronic device, cause the electronic device to: include a high-precision linear module, a laser scanner, a V-shaped block, a pneumatic gripper, a control and data acquisition unit, and an industrial robot; the high-precision linear module is used to control the uniform motion of the shoe body; the laser scanner includes a line laser camera and a line laser base, and the high-precision linear module is disposed with the laser scanner on both sides, which is used to emit high-density line laser and acquire laser profile data; the pneumatic gripper and the V-shaped block are used to fix the shoe body; the control and data acquisition unit includes an embedded controller and a data processing host, which is used to synchronously control the high-precision linear module movement, line laser scanning, and receive data; and the industrial robot is configured with an ink jet head, which is used to perform shoe body line drawing.

[0129] The above description is only the preferred embodiment of the application and the explanation of the applied technical principles. Those skilled in the art should understand that the scope of the application involved in the present application is not limited to the technical solutions formed by the specific combination of the above technical features, and should also cover other technical solutions formed by any combination of the above technical features or their equivalent features without departing from the above inventive concept. For example, the above features can be replaced with the technical features disclosed in the present application (but not limited to) having similar functions to form technical solutions.

[0130] Finally, it should be noted that: the above embodiments are only for illustration and not for limiting the technical solutions of the present application. Although the present application has been described in detail with reference to the above embodiments, those skilled in the art should understand that the present application can still be modified or replaced equivalently without departing from the spirit and scope of the present application. Any modification or partial replacement should be covered in the scope of the claims of the present application.

Claims

1. An intelligent shoe upper side recognition and tracing device, characterized in that, This includes high-precision linear modules, laser scanners, V-blocks, pneumatic grippers, control and data acquisition units, and industrial robots; High-precision linear modules, laser scanners, V-blocks, pneumatic grippers, and industrial robots are all deployed on a vibration damping optical platform; The high-precision linear module is used to control the shoe body to move at a constant speed. The laser scanner is deployed on both sides of the high-precision linear module. The pneumatic gripper and V-shaped baffle are installed above the high-precision linear module. The laser scanner includes a line laser camera and a line laser base. The line laser camera is installed on the line laser base and is used to emit high-density line laser and acquire laser profile data. The control and data acquisition unit is connected to the vibration damping optical platform and includes an embedded controller and a data processing host. It is used to synchronously control the shoe body to perform uniform motion, line laser scanning, and receive point cloud data. The point cloud data is processed using an adaptive growth algorithm and a fracture filling algorithm to obtain a closed curve of the shoe upper side. Based on the closed curve of the shoe upper side, an industrial robot is controlled to complete the tracing of the shoe upper side.

2. The apparatus according to claim 1, characterized in that, The pneumatic gripper and the V-shaped stop are located on the front and rear sides of the shoe body, respectively, and fix the shoe body on the high-precision linear module to move at a constant speed. When the high-precision linear module moves at a constant step size and uniform speed, the laser scanner emits a high-density line laser on the surface of the shoe and acquires the corresponding three-dimensional sampling points of the shoe, thus collecting a full-scale high-density point cloud of the shoe.

3. A method for identifying and tracing the side of a shoe upper using the apparatus as described in any one of claims 1-2, characterized in that, include: S1: The intelligent shoe upper side recognition and drawing device collects point cloud data of the shoe body, performs scalar field transformation, and preprocesses the point cloud data to obtain a set of candidate points for several connected sub-regions and regions to be drawn. S2: The candidate point set of the region to be marked is expanded by an adaptive growth algorithm. Specifically, based on a preset radius, a seed point is found in the KD tree constructed from the candidate point set of the region to be marked. Starting from the seed point, the score of each point in the KD tree is calculated, and the two neighboring points with the highest scores are selected to be added to the next round of growth queue. The growth direction is updated, and the longest connected component is selected as the optimal backbone point by the cumulative length of the projection of the main direction and noise is removed to obtain a continuous backbone point set. S3: The breakpoints in the main point set are filled using a break-filling algorithm. Specifically, the distance from the midpoint to the centroid of two adjacent breakpoints is calculated, and the curvature is selected based on the average radius of the contour. A Bézier peak is set based on the curvature, and quadratic Bézier interpolation is performed to generate an interpolation sequence between the two breakpoints. The normal vector of the interpolation sequence is smoothed using spherical linear interpolation, and the interpolation sequence is then subjected to Gaussian weighted smoothing to obtain the closed curve of the shoe upper side. S4: The industrial robot uses the closed curve of the shoe upper side as the processing path to trace the shoe upper side.

4. The method according to claim 3, characterized in that, The preprocessing of the point cloud data includes assigning attribute values ​​to each point, setting preliminary threshold filtering conditions based on multidimensional attribute values, removing noise points, and extracting connected components based on octrees or the K-nearest neighbor principle.

5. The method according to claim 3, characterized in that, The steps after step S1 and before step S2 include: dividing the discrete set of point cloud data into multiple connected sub-regions, calculating the main direction of the point cloud, evaluating whether the shape of each connected sub-region conforms to the line-drawable feature, using the point cloud data in the connected sub-regions that conform to the line-drawable feature as the candidate point set to be drawn, and filtering the point cloud data that does not conform to the line-drawable feature.

6. The method according to claim 3, characterized in that, The score calculation formula for each point is as follows: in, Indicates the previous growth direction of the current point. This represents the direction vector from the current point to its neighboring points. p neigh p represents a candidate neighbor point. curr Denotes the current growth point, d = ||p neigh -p curr ||, where d represents the Euclidean distance.

7. The method according to claim 3, characterized in that, The formula for updating the growth direction is: in, Indicates the previous growth direction of the current point. This represents the direction vector from the current point to its neighboring points. Indicates the new growth direction at the current point; The formula for calculating the cumulative length of the projection along the principal direction is: Where M represents the total number of points in the connected domain, Pt represents the coordinates of the t-th three-dimensional point in the connected domain after sorting according to the main direction, and Pt+1 represents the next point after sorting according to the main direction that is adjacent to Pt.

8. The method according to claim 3, characterized in that, Project the points in the connected sub-region onto the main direction line, sort them by polar angle and remove outliers. If the distance between adjacent polar angle sorted points is greater than the preset break threshold, then a break exists at that point.

9. The method according to claim 3, characterized in that, When the average radius of the profile is less than the distance from the midpoint of two adjacent break points to the center of mass, the toe curvature is selected as the curvature; otherwise, when the average radius of the profile is greater than the distance from the midpoint of two adjacent break points to the center of mass, the heel curvature is selected as the curvature.

10. The method according to claim 3, characterized in that, The formula for performing quadratic Bezier interpolation is: B(t)=(1-t) 2 s+2(1-t)t p ctrl +t 2 e,t∈[0,1] The formula for calculating the Bezier peak point based on the curvature setting is as follows: p etrl =m-(κL)u The formula for calculating the Gaussian weighting is as follows: Where, p ctrl Let represent the peak point, κ represent the curvature, m represent the peak value, L represent the distance between two fracture points, s and e represent the fracture initiation and fracture termination points, and t represent the interpolation parameters. Let B(t) represent the coordinates of the smoothed interpolation sequence, M represent the total length of the interpolation sequence, k represent the index of the resampled point, and σ represent the standard deviation of the Gaussian function.

11. A computer program product having one or more computer programs thereon, characterized in that, When the computer program is executed by a computer processor, the method described in any one of claims 3-10 is performed.

Images