Sole edge line tracking method based on machine vision

Abstract

The invention relates to the technical field of image processing, and particularly discloses a sole edge line tracking method based on machine vision, which is used for solving the problems of large error and low efficiency of accurate tracking and positioning of a sole edge contour manually in the existing shoemaking process. The three-dimensional contour data of the shoe sole is obtained in real time through the 3D laser scanner to serve as the first contour data, the three-dimensional contour data of the standard shoe sole is selected to serve as the second contour data, and the shoe sole shape difference comparison model is constructed according to the first shape feature and the second shape feature of the shoe sole curved surface to evaluate whether the shoe sole shape is qualified or not. The method comprises the following steps: performing vertical projection on a first sole curved surface model to obtain a sole two-dimensional image, establishing a corresponding relation between adjacent sole edge curves, obtaining a sole offset between the adjacent sole edge curves, and constructing a sole change track model to generate a sole change track so as to realize accurate positioning of sole glue brushing. Dynamic correction and deviation early warning are achieved, and the stability of the production process is guaranteed.

Description

Technical Field

[0001] The present invention relates to the technical field of image processing, and more specifically, to a method for tracking a shoe sole edge line based on machine vision. Background Art

[0002] The shoe production line is a complex automated assembly line, involving multiple processes such as uppers, soles, uppers and lasts. As a critical component of footwear products, the shape, edge contour and processing accuracy of the sole directly determine the wearing comfort, durability and overall appearance of the footwear products. The sole not only needs to meet the requirements of ergonomic design, but also take into account the fashion and functionality of the product. Therefore, the manufacturing process of the sole has put forward high requirements for process accuracy, edge processing and positioning accuracy. In traditional shoemaking technology, due to the limitations of equipment and process means, most of the processes still rely on manual operation. For example, in the placement of the sole, edge line detection, gluing, drying, buckling and other links, there are often certain errors in manual judgment and operation. The uncertainty of manual operation can easily lead to positioning deviations of the sole in subsequent processes, thereby causing a series of process problems, which ultimately affect the overall quality and production efficiency of the product.

[0003] The placement and edge processing of the soles mainly rely on manual vision and experience. Workers need to locate and detect the soles on the assembly line according to predetermined templates or manual markings. However, in the actual production environment, due to the complex and diverse shapes of the soles, slight deformation or deviation may occur during the processing, coupled with the influence of environmental factors such as light and reflection, manual observation is often difficult to ensure the consistency and repeatability of accuracy. Especially in modern shoe factories with high production and high standards, slight operational deviations may cause deviations in the operation of subsequent automated equipment (such as glue coating machines, drying equipment and fastening devices), thereby causing product defects or scrap. Therefore, how to reduce manual operation errors while ensuring efficient production and achieve accurate tracking and positioning of the sole edge contour has become one of the technical problems that the current shoemaking industry needs to solve urgently. In order to solve the above problems, a technical solution is now provided. Summary of the invention

[0004] In order to overcome the above-mentioned defects of the prior art, the present invention provides a sole edge line tracking method based on machine vision, which adopts a sole change trajectory model to generate a sole change trajectory to achieve precise positioning of the sole glue brushing, so as to solve the problem of large error and low efficiency in the existing shoemaking process of manually tracking and positioning the sole edge contour, so as to solve the problems raised in the above-mentioned background technology.

[0005] To achieve the above object, the present invention provides the following technical solutions: A method for tracking the edge line of a shoe sole based on machine vision comprises the following steps: Step 1: Real-time obtain the three-dimensional contour data of the sole through a 3D laser scanner as the first contour data , and select a standard sole to collect its three-dimensional contour data as the second contour data ; Step 2: Generate a first sole surface model by processing the first contour data to extract the first shape feature of the sole, generate a second sole surface model by processing the second contour data to extract the second shape feature of the sole, and construct a sole shape difference comparison model based on the first shape feature and the second shape feature to evaluate whether the sole shape is qualified; Step 3: Obtain a two-dimensional image of the sole by performing a vertical projection on the first sole surface model, extract the first outer edge feature data of the sole based on the two-dimensional image of the sole, perform spatio-temporal positioning on the first outer edge feature data to establish the correspondence between adjacent sole edge curves, and obtain the sole offset between adjacent sole edge curves; Step 4: Cumulatively sum according to the sole offset to construct a sole change trajectory model to generate a sole change trajectory, and achieve precise positioning of sole gluing.

[0006] As a further solution of the present invention, in Step 2, constructing a sole shape difference comparison model based on the first shape feature and the second shape feature to evaluate whether the sole shape is qualified, the specific steps are as follows: Step 21: Perform filtering processing on the first contour data to obtain , generate a first sole surface model through triangular mesh reconstruction ; Extract the first sole surface as a first two-dimensional contour curve through the cross-section projected onto the horizontal plane , where is a normalization parameter, representing the proportion of the position on the first two-dimensional contour curve, is the returned horizontal coordinate corresponding to this position, is the returned vertical coordinate corresponding to this position; Step 22: Perform filtering processing on the second contour data to obtain , generate a second sole surface model through triangular mesh reconstruction ; Extract the second sole surface as a second two-dimensional contour curve through the cross-section projected onto the horizontal plane , where is a normalization parameter, representing the proportion of the position on the second two-dimensional contour curve, is the returned horizontal coordinate corresponding to this position, is the returned vertical coordinate corresponding to this position; Step 23: Based on the first sole surface model and the second sole surface model Calculate the first similarity metric value; the formula for the first similarity metric value is: ; In the formula: is the first similarity metric value, is the height value corresponding to the first sole surface model at the h-th sampling point ; is the height value corresponding to the second sole surface model at the h-th sampling point ; is the total number of sampling points; Step 24, calculate the maximum contour curve deviation value based on the first two-dimensional contour curve and the second two-dimensional contour curve ; the formula for the maximum contour curve deviation value is: ; In the formula: is the maximum contour curve deviation value, is to take the maximum value within the brackets, The point p is taken from the upper boundary of the first two-dimensional contour curve , The point q comes from the lower boundary of the second two-dimensional contour curve , The point p comes from the upper boundary of the second two-dimensional contour curve , The point q is taken from the lower boundary of the first two-dimensional contour curve , is the maximum distance between the upper boundary of the first two-dimensional contour curve and the lower boundary of the second two-dimensional contour curve , is the maximum distance between the upper boundary of the second two-dimensional contour curve and the lower boundary of the first two-dimensional contour curve ; Step 25, import the first similarity metric value and the maximum contour curve deviation value into the sole shape difference comparison model to evaluate whether the sole shape is qualified. The formula for the sole shape difference comparison model is: ; In the formula: is the sole shape difference value, is the first similarity metric value, is the maximum contour curve deviation value, is the curvature difference value between the first two-dimensional contour curve and the second two-dimensional contour curve , is the weight of the first similarity metric value, is the weight of the maximum deviation value of the contour curve, is the weight of the curvature difference value; Step 26: Compare the sole shape difference value with a preset sole shape difference threshold. If the sole shape difference value is greater than or equal to the preset sole shape difference threshold, the sole shape is unqualified; if the sole shape difference value is less than the preset sole shape difference threshold, the sole shape is qualified.

[0007] As a further solution of the present invention, in step 3, a sole two-dimensional image is obtained by vertically projecting the first sole curved surface model. Based on the sole two-dimensional image, the first outer edge feature data of the sole is extracted. The corresponding relationship between adjacent sole edge curves is established by spatio-temporal positioning of the first outer edge feature data, and the sole offset amount between adjacent sole edge curves is obtained. The specific steps are as follows: Step 31: Obtain a sole two-dimensional image by vertically projecting the first sole curved surface model , extract the sole edge binary image based on image processing , and process the sole edge binary image using connected component analysis and morphological operations to ensure the continuity of the sole edge line, and extract each sole edge contour line , where b is the b-th sole; Step 32: Aggregate all sole edge lines to obtain an edge contour dataset , where is the b-th sole, and B is the number of sole edge line contours in the edge contour dataset; Step 33: For two consecutive sole edge lines and , calculate the matching correspondence between the two sole edge lines using feature point matching. The sole edge line matching error function is: ; In the formula: is the b-th sole edge line, is the (b + 1)-th sole edge line, is the normalization parameter, representing the proportion of the position on the (b + 1)-th sole edge line; Step 34: For the matched corresponding points and , the local offset is: ; In the formula: is the coordinate of the corresponding point determined by feature point matching on the b-th sole edge line , It is the (b + 1)-th sole edge line The coordinates of the corresponding points determined by feature point matching on it.

[0008] As a further solution of the present invention, in step 4, according to the cumulative summation of the sole offset, a sole change trajectory model is constructed to generate a sole change trajectory to achieve precise positioning of sole gluing. The formula of the sole change trajectory model is: ; In the formula: is the output of the sole change trajectory model, is the initial local offset, is the local offset of the b-th sole edge line, and B is the number of sole edge line contours in the edge contour dataset.

[0009] The technical effects and advantages of a method for tracking the sole edge line based on machine vision in the present invention: The present invention uses a 3D laser scanner to obtain the three-dimensional contour data of the sole in real time as the first contour data, and selects a standard sole to collect its three-dimensional contour data as the second contour data. According to the first shape feature and the second shape feature of the sole surface, a sole shape difference comparison model is constructed to evaluate whether the sole shape is qualified. Through shape feature extraction and standard comparison, automatic evaluation of sole qualification is realized, reducing manual judgment errors; by performing vertical projection on the first sole surface model to obtain a 2D image of the sole, establishing the corresponding relationship between adjacent sole edge curves, and obtaining the sole offset between adjacent sole edge curves, providing an accurate basis for generating the change trajectory, and then constructing a sole change trajectory model to generate a sole change trajectory to achieve precise positioning of sole gluing, thereby guiding the gluing equipment to achieve precise positioning, ensuring the perfect docking of the vamp and the sole, realizing dynamic correction and deviation warning, and ensuring the stability of the production process. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] Figure 1 is a schematic flow chart of a method for tracking the sole edge line based on machine vision provided by the present invention; Figure 2 is a schematic flow chart of step 2 in a method for tracking the sole edge line based on machine vision provided by the present invention; Figure 3 is a two-dimensional contour comparison analysis diagram provided by the present invention; Figure 4 is a sole shape difference heat map and index statistical chart provided by the present invention; Figure 5 is a schematic diagram of a 3D sole surface model provided by the present invention; Figure 6 is a sole edge trajectory analysis diagram provided by the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0011] The following will clearly and completely describe the technical solutions in the present invention in conjunction with the accompanying drawings in the present invention. Obviously, the described technical solutions are only a part of the present invention, rather than all of it. All other technical solutions obtained by those of ordinary skill in the art based on the technical solutions in the present invention without creative efforts fall within the scope of protection of the present invention.

[0012] Embodiment 1 Figure 1 is a schematic flowchart of a method for tracking the sole edge line based on machine vision provided by the present invention. As Figure 1 shown, a method for tracking the sole edge line based on machine vision includes the following steps: Step 1: Real-time obtain the three-dimensional contour data of the sole through a 3D laser scanner as the first contour data , and select a standard sole to collect its three-dimensional contour data as the second contour data ; Step 2: Process the first contour data to generate a first sole surface model, extract the first shape feature of the sole, process the second contour data to generate a second sole surface model, extract the second shape feature of the sole, and construct a sole shape difference comparison model based on the first shape feature and the second shape feature to evaluate whether the sole shape is qualified; Step 3: Perform a vertical projection on the first sole surface model to obtain a two-dimensional image of the sole, extract the first outer edge feature data of the sole based on the two-dimensional image of the sole, perform spatio-temporal positioning on the first outer edge feature data to establish the correspondence between adjacent sole edge curves, and obtain the sole offset between adjacent sole edge curves; Step 4: Cumulatively sum according to the sole offset to construct a sole change trajectory model to generate a sole change trajectory, and achieve precise positioning of sole gluing.

[0013] Collect the three-dimensional contour data of the sole and the standard sole using a 3D laser scanner, which can accurately capture the minute deformations and details of the sole; process the data to construct a sole surface model and extract shape features, and then establish a shape difference comparison model to evaluate whether the sole is qualified. Through shape feature extraction and standard comparison, the automatic evaluation of the sole's qualification is realized, reducing manual judgment errors; project the first sole surface model vertically to obtain a two-dimensional image, extract the outer edge data of the sole from it, and use the DTW algorithm to achieve the spatio-temporal positioning and matching of continuous edges. The spatio-temporal positioning and matching algorithm enables the capture of subtle offsets between continuous sole edges, providing an accurate basis for generating the change trajectory; according to the locally obtained offset amount from the matching, adopt the cumulative and smoothing method to construct a sole change trajectory model, providing real-time feedback data for gluing positioning. The change trajectory model generated by cumulative summation can dynamically reflect the displacement of the sole during the production process, thereby guiding the gluing equipment to achieve precise positioning, ensuring the perfect docking of the vamp and the sole, and feeding the trajectory data back to the automated control system to achieve dynamic correction and deviation warning, ensuring the stability of the production process and product consistency.

[0014] Specifically, as Figure 2 shown in the flow diagram of step 2 in a method for tracking the sole edge line based on machine vision, in step 2, construct a sole shape difference comparison model based on the first shape feature and the second shape feature to evaluate whether the sole shape is qualified. The specific steps are as follows: Step 21, perform filtering processing on the first contour data to obtain , generate the first sole surface model through triangular mesh reconstruction ; extract the first sole surface as the first two-dimensional contour curve through the cross-section projected onto the horizontal plane , where is the normalization parameter, representing the proportion of the position on the first two-dimensional contour curve, returns the corresponding horizontal coordinate at this position, returns the corresponding vertical coordinate at this position; Step 22, perform filtering processing on the second contour data to obtain , generate the second sole surface model through triangular mesh reconstruction ; extract the second sole surface as the second two-dimensional contour curve through the cross-section projected onto the horizontal plane , where is the normalization parameter, representing the proportion of the position on the second two-dimensional contour curve, returns the corresponding horizontal coordinate at this position, returns the corresponding vertical coordinate at this position; Step 23, based on the first sole surface model and the second sole surface model Calculate the first similarity metric value; the formula for the first similarity metric value is: ; In the formula: is the first similarity metric value, is the height value corresponding to the first sole surface model at the h-th sampling point ; is the height value corresponding to the second sole surface model at the h-th sampling point ; is the total number of sampling points; it should be noted that and represent the sampling points of the two models at the same position respectively; Step 24, based on the first two-dimensional contour curve and the second two-dimensional contour curve calculate the maximum contour curve deviation value; the formula for the maximum contour curve deviation value is: ; In the formula: is the maximum contour curve deviation value, is to take the maximum value within the brackets, the point p is taken from the upper boundary of the first two-dimensional contour curve , the point q comes from the lower boundary of the second two-dimensional contour curve , the point p comes from the upper boundary of the second two-dimensional contour curve , the point q is taken from the lower boundary of the first two-dimensional contour curve , is the maximum distance between the upper boundary of the first two-dimensional contour curve and the lower boundary of the second two-dimensional contour curve , is the maximum distance between the upper boundary of the second two-dimensional contour curve and the lower boundary of the first two-dimensional contour curve ; Step 25, import the first similarity metric value and the maximum contour curve deviation value into the sole shape difference comparison model to evaluate whether the sole shape is qualified. The formula for the sole shape difference comparison model is: ; In the formula: is the sole shape difference value, is the first similarity metric value, is the maximum contour curve deviation value, is the first two-dimensional contour curve and the second two-dimensional contour curve of the curvature difference value, is the weight of the first similarity metric value, is the weight of the maximum deviation value of the contour curve, is the weight of the curvature difference value; Step 26, compare the sole shape difference value with the preset sole shape difference threshold. If the sole shape difference value is greater than or equal to the preset sole shape difference threshold, the sole shape is unqualified; if the sole shape difference value is less than the preset sole shape difference threshold, the sole shape is qualified.

[0015] Such as Figure 5 The two-dimensional contour comparison analysis diagram shown; it shows the sole surface model reconstructed by three-dimensional modeling or 3D scanning, and can intuitively see the overall shape, pattern and concave-convex structure of the sole. The length information marked in the figure is used to reference the size range of the sole from a certain perspective, providing basic data for subsequent shape difference analysis.

[0016] Such as Figure 4 The sole shape difference heat map and index statistical chart shown, which shows two contour curves (usually one is the standard sole contour and the other is the current sole contour) obtained after projecting the sole surface onto a two-dimensional plane. The relative position difference between the green curve and the blue curve intuitively shows the offset and deformation of the two in the X and Y directions. The horizontal axis in the figure is the X value (mm), and the vertical axis is the Y value (mm), which can be used to evaluate the difference between the current sole and the standard contour.

[0017] Such as Figure 3 The schematic diagram of the 3D sole surface model shown, which contains a matrix diagram, uses the depth of color or the size of the value to represent the difference degree of different sampling points / regions, and gives multiple shape difference measurement indexes; similarity metric (d SIM ): The value is 0.892, representing the root mean square error or similarity of the surface height difference; maximum deviation value (H(C(t),C2(t2))): 2.45mm, reflecting the maximum offset of the two two-dimensional contour curves at the most extreme positions; curvature difference (E cu ): 0.156, used to measure the local curvature difference of the two contour curves; total shape difference value (E TOT ): 0.248, which synthesizes the above multiple indexes with weights and is used to overall evaluate the closeness of the sole to the standard shape.

[0018] The three-dimensional contour data collected by 3D laser scanning technology can accurately reflect the minute deformations of the shoe sole. The surface model obtained through filtering and reconstruction can truly restore the surface details of the shoe sole. The model is compared from two perspectives: on the one hand, the similarity metric calculated through height values (surface function) can reflect the height differences in the overall shape; on the other hand, through the maximum deviation and curvature differences of the two-dimensional contour curves, local contour changes and detail deviations can be captured. The combination of the two makes the shape comparison more comprehensive. By constructing a comprehensive difference index and introducing a weight coefficient, the influence degree of each index can be flexibly adjusted according to different production requirements, realizing the quantitative evaluation of the qualifiedness of the shoe sole shape and avoiding the one-sidedness that may be brought by a single index. The model constructed based on machine vision and mathematical models can automatically collect, process, and evaluate the shoe sole shape data on the production line, improving the detection speed and consistency, and reducing manual intervention and the errors it brings. Each index in the model (height difference, contour deviation, curvature difference) can be expanded and modified according to actual application needs, applicable to the detection of different shoe types and different process requirements, providing a unified platform for intelligent shoemaking. Using the multi-dimensional comparison of three-dimensional contour data and two-dimensional contour curves, the precise quantitative evaluation of the shoe sole shape is realized, providing data support for subsequent process control. At the same time, it has the advantages of high precision, automation, and flexible expansion, and can effectively improve the quality and efficiency of the shoemaking production line.

[0019] Specifically, the first two-dimensional contour curve and the second two-dimensional contour curve The calculation formula for the curvature difference value is: ; In the formula: is the curvature difference value between the first two-dimensional contour curve and the second two-dimensional contour curve , is the total arc length of the second two-dimensional contour curve , is the curvature of the first two-dimensional contour curve , is the curvature of the second two-dimensional contour curve .

[0020] Specifically, in step 3, the shoe sole two-dimensional image is obtained by vertically projecting the first shoe sole surface model. Based on the shoe sole two-dimensional image, the first outer edge feature data of the shoe sole is extracted. The corresponding relationship between adjacent shoe sole edge curves is established through spatio-temporal positioning of the first outer edge feature data, and the shoe sole offset between adjacent shoe sole edge curves is obtained. The specific steps are as follows: Step 31, the shoe sole two-dimensional image is obtained by vertically projecting the first shoe sole surface model , and the shoe sole edge binary image is extracted based on image processing , the binary image of the sole edge is processed by connected component analysis and morphological operations to ensure the continuity of the sole edge line, and each sole edge contour line is extracted , where b is the b-th sole; Step 32, summarize all the sole edge lines to obtain an edge contour dataset , where is the b-th sole, and B is the number of sole edge line contours in the edge contour dataset; Step 33, for two consecutive sole edge lines and , the matching correspondence between the two sole edge lines is calculated based on feature point matching, and the sole edge line matching error function is: ; In the formula: is the b-th sole edge line, is the (b + 1)-th sole edge line, is the normalization parameter, representing the ratio of the position on the (b + 1)-th sole edge line; Step 34, for the matched corresponding points and , the local offset is: ; In the formula: is the coordinate of the corresponding point determined by feature point matching on the b-th sole edge line , is the coordinate of the corresponding point determined by feature point matching on the (b + 1)-th sole edge line .

[0021] Generate a two-dimensional image of the sole through vertical projection using the first sole surface model, extract the first outer edge feature data of the sole based on image processing technology, then extract the edge contours of each sole to form an edge contour dataset, and then use the feature point matching method to establish a corresponding relationship between the edge curves of consecutive soles and calculate the local offset; convert the three-dimensional sole surface into a two-dimensional image through vertical projection, and use mature image processing algorithms (such as Canny edge detection and morphological operations) to obtain stable and continuous outer edge data of the sole, so as to ensure the accuracy of subsequent matching; use connected component analysis and morphological operations to effectively fill the edge fracture area, ensure that the extracted contour curve has good continuity and integrity, and reduce the matching error caused by edge missing; use the feature point-based matching method (such as the DTW algorithm) to perform spatio-temporal positioning on the adjacent sole edges, establish an accurate corresponding relationship, and can accurately capture the subtle offsets generated by the placement, vibration or micro-deformation of the sole during the production process; by calculating the local offset, the deformation information between consecutive soles can be quantified, providing accurate data support for the subsequent generation of the sole change trajectory model and dynamic correction; collect the sole edge data in real time and calculate the offset between adjacent soles, providing an on-line correction basis for the gluing process, which helps to achieve high-precision and stable automated production.

[0022] Specifically, in step 4, construct a sole change trajectory model according to the cumulative sum of the sole offsets to generate the sole change trajectory and realize the accurate positioning of sole gluing. The formula of the sole change trajectory model is: ; In the formula: is the output of the sole change trajectory model, is the initial local offset, is the local offset of the b-th sole edge line, and B is the number of sole edge line contours in the edge contour dataset.

[0023] Such as Figure 6 The sole edge trajectory analysis diagram shown demonstrates the curves of two edge trajectories evolving over time, where the abscissa is time (seconds) and the ordinate is the offset (millimeters). The blue curve and the green curve represent the motion / deformation conditions of two compared sole edges (or the same edge in different coordinate directions) respectively. Both trajectories show periodic changes similar to sine waves, indicating that the edges repeat similar motion patterns within a certain time interval; the blue curve and the green curve are not synchronized at the peak and trough positions, reflecting the time phase difference between the two edges (or two directions); the numerical sizes of the peaks and troughs can be used to evaluate the amplitude of the edge displacement, thereby judging the offset or deformation degree of the sole during this time period.

[0024] By accumulating the local offsets between consecutive sole edges, a continuous change trajectory can be formed, which can reflect the displacement and deformation of the sole on the production line in real time. The change trajectory provides accurate real-time positioning information for the gluing equipment, enabling the glue brushing device to perform dynamic correction according to the actual position of the sole and achieve precise alignment. During the production process, the sole may be affected by minor placement errors or vibrations. Cumulative summation can capture and accumulate these offset information, so as to compensate for the position during gluing and ensure seamless docking of the vamp and the sole. Through the vectorized local offset data, the generated change trajectory can provide more stable position information than single-frame detection, providing a high-precision positioning basis for the glue brushing process. The generated sole change trajectory is not only used for immediate positioning but also as a basis for subsequent data analysis. By analyzing the trend of the trajectory change, it is possible to determine whether there are systematic deviations in the production process, thereby guiding the optimization and adjustment of process parameters. A closed-loop feedback system is formed between the change trajectory model and the gluing equipment to ensure that the system can automatically adjust when deviations occur, improving the overall product consistency and quality.

[0025] Embodiment 2 Regarding the above step 3, a sole two-dimensional image is obtained by vertically projecting the first sole surface model. Based on the sole two-dimensional image, the first outer edge feature data of the sole is extracted. The spatio-temporal positioning of the first outer edge feature data is performed to establish the correspondence between adjacent sole edge curves, and the sole offset between adjacent sole edge curves is obtained. The following is an example for illustration: During the process of the shoe-making production line, the contour data of three soles are collected, and the surface data of the first sole, the second sole, and the third sole are generated through 3D scanning. These data are processed through the following steps: At this time, the vertical projection of the first sole surface model has been completed, and a two-dimensional image is obtained. The binary image of the first sole edge is extracted through image processing, and connected component analysis and morphological operations are used to ensure the continuity of the edge line. Finally, the edge contour lines of each sole are extracted, and the first sole edge contour curve data extracted from the image are given as shown in Table 1:

[0026] The edge contour curves of the second and third soles are extracted from their surface data using the same method. The data structure of each curve is the same as that of the first sole edge contour line, and the second sole edge contour curve data extracted from the image are given as shown in Table 2:

[0027] The third sole edge contour curve data extracted from the image are shown in Table 3:

[0028] Based on the above table data, 3 soles are summarized. The contour curve of each sole is sampled into several points, which serves as the basis for subsequent analysis.

[0029] By using the method based on feature point matching, calculate the matching error between adjacent sole edge lines. For sole 1 and sole 2, we find their corresponding points at each position t through feature point matching , and calculate the error between each pair of corresponding points. At t = 0.5t, the corresponding point of the first sole is , and the corresponding point of the second sole is , then , continue to calculate the matching error for all corresponding points at t until a complete error matrix is obtained.

[0030] After feature point matching, we obtain the set of matching corresponding points and , calculate the local offset of each pair of corresponding points. For example, when t = 5, the corresponding points of the first sole and the second sole are respectively: ; ; At this time, the local offset is: ; Calculate the local offset for each pair of corresponding points based on the above calculation method, and construct the entire change trajectory.

[0031] In the embodiment of the present invention, the three-dimensional contour data of the sole is obtained in real time by a 3D laser scanner as the first contour data, and the three-dimensional contour data of a standard sole is collected as the second contour data. A sole shape difference comparison model is constructed according to the first shape feature and the second shape feature of the sole surface to evaluate whether the sole shape is qualified. Through shape feature extraction and standard comparison, automatic evaluation of sole qualification is realized, reducing manual judgment errors; by performing a vertical projection on the first sole surface model to obtain a two-dimensional image of the sole, establishing the corresponding relationship between adjacent sole edge curves, and obtaining the sole offset between adjacent sole edge curves, providing an accurate basis for generating the change trajectory, and then constructing a sole change trajectory model to generate the sole change trajectory to achieve precise positioning of sole gluing, thereby guiding the gluing equipment to achieve precise positioning, ensuring the perfect docking of the vamp and the sole, realizing dynamic correction and deviation warning, and ensuring the stability of the production process.

[0032] The above is only the specific implementation manner of this application, but the protection scope of this application is not limited thereto. Any person skilled in the art within the technical scope disclosed by this application can easily think of changes or substitutions, which should all be covered within the protection scope of this application. Therefore, the protection scope of this application should be subject to the protection scope of the claimed rights.

[0033] Finally, the above are only the preferred solutions of the present invention and are not used to limit the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present invention shall be included within the protection scope of the present invention.

Claims

1. A method for tracking the edge line of a sole based on machine vision, characterized in that, It includes the following steps: Step 1, obtain the three-dimensional contour data of the sole in real time through a 3D laser scanner as the first contour data , and select a standard sole to collect its three-dimensional contour data as the second contour data ; Step 2: Generate a first sole surface model by processing the first contour data to extract the first shape feature of the sole, generate a second sole surface model by processing the second contour data to extract the second shape feature of the sole, and construct a sole shape difference comparison model based on the first shape feature and the second shape feature to evaluate whether the sole shape is qualified; Step 3: Obtain a two-dimensional sole image by vertically projecting the first sole surface model, extract the first outer edge feature data of the sole based on the two-dimensional sole image, perform spatio-temporal positioning on the first outer edge feature data to establish the correspondence between adjacent sole edge curves, and obtain the sole offset between adjacent sole edge curves; Step 4: Construct a sole change trajectory model by cumulative summation of the sole offset to generate a sole change trajectory, and achieve precise positioning of sole gluing.

2. The method for tracking the sole edge line based on machine vision according to claim 1, wherein In Step 2, constructing a sole shape difference comparison model based on the first shape feature and the second shape feature to evaluate whether the sole shape is qualified, the specific steps are as follows: Step 21, perform filtering on the first contour data to obtain , and generate a first sole surface model through triangular mesh reconstruction ; extract the first sole surface as a first two-dimensional contour curve through the cross-section projected onto the horizontal plane , where is a normalization parameter representing the proportion of the position on the first two-dimensional contour curve, returns the corresponding horizontal coordinate at this position, returns the corresponding vertical coordinate at this position; Step 22, for the second contour data perform filtering to obtain , and generate a second sole surface model through triangular mesh reconstruction ; extract a second two-dimensional contour curve from the section where the second sole surface is projected onto the horizontal plane , where is a normalization parameter representing the proportion of the position on the second two-dimensional contour curve, returns the corresponding horizontal coordinate at this position, returns the corresponding vertical coordinate at this position; Step 23, based on the first sole surface model and the second sole surface model calculate a first similarity metric value; the formula for the first similarity metric value is: ； Wherein: is the first similarity metric value, is the height value corresponding to the first sole surface model at the h-th sampling point therein, is the height value corresponding to the second sole surface model at the h-th sampling point therein, is the total number of sampling points; Step 24, based on the first two-dimensional contour curve and the second two-dimensional contour curve calculate the maximum deviation value of the contour curve; the calculation formula for the maximum deviation value of the contour curve is: ； In the formula: is the maximum deviation value of the contour curve, is to take the maximum value within the brackets, means that point p is taken from the upper boundary of the first two-dimensional contour curve ; means that point q comes from the lower boundary of the second two-dimensional contour curve ; means that point p comes from the upper boundary of the second two-dimensional contour curve ; means that point q is taken from the lower boundary of the first two-dimensional contour curve ; is the upper boundary of the first two-dimensional contour curve and the maximum distance between the lower boundary of the second two-dimensional contour curve ; is the upper boundary of the second two-dimensional contour curve and the maximum distance between the lower boundary of the first two-dimensional contour curve . Step 25: Import the first similarity metric value and the maximum deviation value of the contour curve into the sole shape difference comparison model to evaluate whether the sole shape is qualified. The formula of the sole shape difference comparison model is: ； Wherein: is the sole shape difference value, is the first similarity metric value, is the maximum deviation value of the contour curve, is the first two-dimensional contour curve and the second two-dimensional contour curve is the curvature difference value, is the weight of the first similarity metric value, is the weight of the maximum deviation value of the contour curve, is the weight of the curvature difference value; Step 26: Compare the sole shape difference value with the preset sole shape difference threshold. If the sole shape difference value is greater than or equal to the preset sole shape difference threshold, the sole shape is unqualified; if the sole shape difference value is less than the preset sole shape difference threshold, the sole shape is qualified.

3. The method for tracking the sole edge line based on machine vision according to claim 1, wherein, In Step 3, obtaining a two-dimensional sole image by vertically projecting the first sole surface model, extracting the first outer edge feature data of the sole based on the two-dimensional sole image, performing spatio-temporal positioning on the first outer edge feature data to establish the correspondence between adjacent sole edge curves, and obtaining the sole offset between adjacent sole edge curves, the specific steps are as follows: Step 31: Obtain a two-dimensional image of the sole by performing a vertical projection on the first sole curved surface model , extract a binary image of the sole edge based on image processing , and perform connected component analysis and morphological operations on the binary image of the sole edge to ensure the continuity of the sole edge line and extract each sole edge contour line , where b is the b-th sole; Step 32, summarize all the sole edge lines to obtain an edge contour dataset , where is the b-th sole, and B is the number of sole edge line contours in the edge contour dataset; Step 33, for two consecutive sole edge lines and , calculate the matching correspondence between the two sole edge lines by feature point matching. The sole edge line matching error function is as follows: ； In the formula: is the b-th sole edge line, is the (b + 1)-th sole edge line, is a normalization parameter, representing the proportion of the position on the (b + 1)-th sole edge line; Step 34, for the corresponding points after matching and , the local offset is as follows: ； Wherein: is the coordinate of the corresponding point determined by feature point matching on the b-th sole edge line , and is the coordinate of the corresponding point determined by feature point matching on the (b + 1)-th sole edge line .

4. The method for tracking the sole edge line based on machine vision according to claim 1, wherein In Step 4, constructing a sole change trajectory model by cumulative summation of the sole offset to generate a sole change trajectory to achieve precise positioning of sole gluing. The formula of the sole change trajectory model is: ； In the formula: is the output of the sole change trajectory model, is the initial local offset, is the local offset of the b-th sole edge line, and B is the number of sole edge line contours in the edge contour dataset.

Images