阅读说明：本技术 一种定制鞋履的方法 (Method for customizing shoe ) 是由 潘少宝 陈标 于 2021-08-31 设计创作，主要内容包括：本发明公开了一种定制鞋履的方法,包括：建立鞋子三维模型,并将鞋子三维模型导入至鞋子定制平台进行渲染；建立鞋楦数据获取模型,并将鞋楦数据获取模型导入至鞋子定制平台；建立脚型数据获取模型,并将脚型数据获取模型导入至鞋子定制平台；获取顾客的定制信息和脚型测量信息；对顾客的脚型测量信息进行预处理,并输入至脚型数据获取模型得到顾客的脚型数据；将顾客的脚型数据输入至鞋楦数据获取模型得到鞋楦数据；根据鞋楦数据制作鞋楦,并根据制作好的鞋楦制作鞋子,本发明可使客户参与到设计过程中,同时,顾客只需发送脚的视频信息,提高了顾客的便利性与满意度。(The invention discloses a method for customizing shoes, which comprises the following steps: building a shoe three-dimensional model, and guiding the shoe three-dimensional model into a shoe customization platform for rendering; establishing a shoe tree data acquisition model, and importing the shoe tree data acquisition model to a shoe customization platform; establishing a foot type data acquisition model, and importing the foot type data acquisition model into a shoe customization platform; obtaining customized information and foot shape measurement information of a customer; preprocessing the foot shape measurement information of the customer, and inputting the foot shape measurement information into a foot shape data acquisition model to obtain foot shape data of the customer; inputting the foot shape data of the customer into a shoe tree data acquisition model to obtain shoe tree data; the invention can enable the customer to participate in the design process, and meanwhile, the customer only needs to send the video information of the foot, thereby improving the convenience and the satisfaction of the customer.)

1. A method of customizing footwear, comprising:

building a shoe three-dimensional model, and guiding the shoe three-dimensional model into a shoe customization platform for rendering;

establishing a shoe tree data acquisition model, and importing the shoe tree data acquisition model to a shoe customization platform;

establishing a foot type data acquisition model, and importing the foot type data acquisition model into a shoe customization platform;

obtaining customized information and foot shape measurement information of a customer;

preprocessing the foot shape measurement information of the customer, and inputting the foot shape measurement information into a foot shape data acquisition model to obtain foot shape data of the customer;

inputting the foot shape data of the customer into a shoe tree data acquisition model to obtain shoe tree data;

and manufacturing a shoe tree according to the shoe tree data, and manufacturing shoes according to the manufactured shoe tree.

2. The method of customizing footwear according to claim 1, wherein the step of building a three-dimensional model of a shoe comprises the steps of: establishing a contour part model of the shoe through Maya three-dimensional software; and (4) carrying out detail adjustment on the established shoe by a subdivision modeling method to obtain a complete shoe three-dimensional model.

3. The method of customizing footwear according to claim 1, wherein importing the shoe three-dimensional model into the shoe customization platform for rendering comprises rendering the shoe three-dimensional model data by texture, the rendering the shoe three-dimensional model data by texture comprising: storing the three-dimensional shoe model data and rendering texture resources; carrying out coordinate transformation processing on the vertexes of the shoe three-dimensional model through a vertex shader, and carrying out cutting processing on the shoe three-dimensional model subjected to the coordinate transformation processing to obtain primitive coordinates of the shoe three-dimensional model in a cutting space; converting each primitive coordinate of the shoe three-dimensional model in the cutting space to a screen coordinate system, and performing rasterization processing on the primitive in the screen coordinate system to obtain a fragment; texture sampling is carried out through a fragment shader to obtain texture coordinates of each fragment, and the color of each fragment is calculated through the texture coordinates of each fragment; carrying out template test and depth test on each fragment, carrying out fusion processing on the color of the template test and depth test fragment and the stored color, and storing the fragment subjected to the color fusion processing; and acquiring the three-dimensional model of the shoe after texture rendering.

4. The method of customizing a footwear according to claim 3, wherein importing the shoe three-dimensional model into the shoe customization platform for rendering further comprises rendering the shoe three-dimensional model data by a map, the rendering the shoe three-dimensional model data by the map comprising the steps of: storing the rendering map resources, and storing and packaging the shoe three-dimensional model; mapping the rendering map and each part in the three-dimensional model of the shoe; and obtaining a three-dimensional model of the shoe after the mapping rendering.

5. The method of customizing footwear according to claim 1, wherein said establishing a last data acquisition model comprises the steps of: acquiring a matching relation database of the foot shape and the indexes of the shoe tree in length, width and circumference; constructing a first convolution neural network model, inputting data in the matching relation database into the first convolution neural network model for training to obtain a preliminary shoe tree data acquisition model; acquiring foot shape sample data of a tester and comfort information of the tester trying on a shoe tree corresponding to the foot shape; adjusting the shoe tree corresponding to the foot shape of the tester according to the comfort degree information until the tester feels comfortable; inputting the foot type sample data of the tester and the corresponding shoe tree data into a matching relation database to update the data in the matching relation database; and inputting the data in the matching relation database into the preliminary shoe tree data acquisition model for training to obtain a final shoe tree data acquisition model.

6. The method of customizing footwear according to claim 5, wherein said obtaining a database of matching foot shapes to last in length, width, circumference metrics comprises the steps of: obtaining the matching relation between the foot shape and the shoe tree in the length index according to the relation between the shoe tree bottom sample length and the foot length data, the back tolerance and the allowance; obtaining the matching relation between the foot shape and the width index of the shoe tree according to the relation between the width of the shoe tree, the foot width data and the allowance; obtaining the matching relation between the foot shape and the shoe tree in the circumference index according to the relation between the shoe tree circumference and the foot circumference data and the allowance; and storing the matching relations to obtain a matching relation database.

7. The method of customizing footwear according to claim 1, wherein said establishing a foot-type data acquisition model comprises the steps of: acquiring a plurality of foot images through web crawlers and camera shooting; making a data set by using the obtained foot images, and dividing the data set into a training set, a testing set and a verification set; inputting the training set into the established second convolutional neural network model for training to obtain a preliminary foot shape data acquisition model; and fine-tuning the preliminary foot type data acquisition model by a fine-tune method, modifying related parameters and acquiring a final foot type data acquisition model.

8. The method of customizing footwear according to claim 7, wherein the step of using the captured foot images to produce a data set and dividing the data set into a training set, a testing set, and a verification set comprises the steps of: marking the obtained foot image by a LabelImg image marking tool; scaling the marked foot images by a bilinear interpolation algorithm to enable the sizes of all the images to be preset values, and storing the scaled foot images; converting the foot image after the zooming processing from an RGB space to an HSV space through a brightness conversion formula, and turning over the foot image after the zooming processing; storing the foot images after brightness conversion and turnover processing; normalizing the stored foot images by a linear function normalization method to obtain a data set; the data set is divided into a training set, a testing set and a verification set according to a preset proportion.

9. The method of customizing footwear according to claim 1, wherein said obtaining customer customization information and foot measurement information comprises the steps of: the customer sends shoe customization information to the shoe customization platform; the method comprises the following steps that a manufacturer obtains contact information of a customer and information of shoes designed by the customer independently on a shoe customization platform; acquiring video information of a customer foot through a camera; the customer sends the video information of the foot to the shoe customization platform.

10. The method of customizing footwear according to claim 1, wherein said preprocessing the foot measurement information of the customer comprises the steps of: acquiring a video file of a customer foot, and acquiring the total frame number of the video file; carrying out image correction processing on each frame image of the video through Hough transformation; starting from a first frame image of a video, acquiring an image difference value of a current frame image and a next frame image; and taking the current frame image with the image difference value larger than the preset threshold value as a key frame image for storage, and removing the current frame image with the image difference value larger than the preset threshold value.

Technical Field

The invention relates to the technical field of intelligent customization, in particular to a method for customizing shoes.

Background

At present, manufacturers design and produce shoes according to popular elements and the development trend of the market, however, for some shoes full of precious goods on a sales platform, a large amount of time is also spent for a customer to select a pair of shoes which the customer likes, so that the customer satisfaction is reduced, and therefore, it is necessary for the customer to participate in the design process, and meanwhile, it is necessary for the customer to intelligently obtain foot type data of the customer and make shoe trees and shoes according to the foot type data.

Therefore, it is an urgent need to solve the above problems by providing a new technical solution.

Disclosure of Invention

In view of the above, the present invention provides a method for customizing a shoe, so as to solve the above technical problems.

In order to achieve the purpose, the invention provides the following technical scheme:

a method of customizing footwear, comprising: building a shoe three-dimensional model, and guiding the shoe three-dimensional model into a shoe customization platform for rendering; establishing a shoe tree data acquisition model, and importing the shoe tree data acquisition model to a shoe customization platform; establishing a foot type data acquisition model, and importing the foot type data acquisition model into a shoe customization platform; obtaining customized information and foot shape measurement information of a customer; preprocessing the foot shape measurement information of the customer, and inputting the foot shape measurement information into a foot shape data acquisition model to obtain foot shape data of the customer; inputting the foot shape data of the customer into a shoe tree data acquisition model to obtain shoe tree data; and manufacturing a shoe tree according to the shoe tree data, and manufacturing shoes according to the manufactured shoe tree.

In the above scheme, the building of the three-dimensional model of the shoe comprises the following steps: establishing a contour part model of the shoe through Maya three-dimensional software; and (4) carrying out detail adjustment on the established shoe by a subdivision modeling method to obtain a complete shoe three-dimensional model.

In the above scheme, the step of importing the shoe three-dimensional model into the shoe customization platform for rendering includes rendering the shoe three-dimensional model data by using textures, and the step of rendering the shoe three-dimensional model data by using the textures includes the following steps: storing the three-dimensional shoe model data and rendering texture resources; carrying out coordinate transformation processing on the vertexes of the shoe three-dimensional model through a vertex shader, and carrying out cutting processing on the shoe three-dimensional model subjected to the coordinate transformation processing to obtain primitive coordinates of the shoe three-dimensional model in a cutting space; converting each primitive coordinate of the shoe three-dimensional model in the cutting space to a screen coordinate system, and performing rasterization processing on the primitive in the screen coordinate system to obtain a fragment; texture sampling is carried out through a fragment shader to obtain texture coordinates of each fragment, and the color of each fragment is calculated through the texture coordinates of each fragment; carrying out template test and depth test on each fragment, carrying out fusion processing on the color of the template test and depth test fragment and the stored color, and storing the fragment subjected to the color fusion processing; and acquiring the three-dimensional model of the shoe after texture rendering.

In the above scheme, the importing the shoe three-dimensional model into the shoe customization platform for rendering further includes rendering the shoe three-dimensional model data by using a map, and the rendering the shoe three-dimensional model data by using the map includes the following steps: storing the rendering map resources, and storing and packaging the shoe three-dimensional model; mapping the rendering map and each part in the three-dimensional model of the shoe; and obtaining a three-dimensional model of the shoe after the mapping rendering.

In the above scheme, the building of the last data obtaining model includes the following steps: acquiring a matching relation database of the foot shape and the indexes of the shoe tree in length, width and circumference; constructing a first convolution neural network model, inputting data in the matching relation database into the first convolution neural network model for training to obtain a preliminary shoe tree data acquisition model; acquiring foot shape sample data of a tester and comfort information of the tester trying on a shoe tree corresponding to the foot shape; adjusting the shoe tree corresponding to the foot shape of the tester according to the comfort degree information until the tester feels comfortable; inputting the foot type sample data of the tester and the corresponding shoe tree data into a matching relation database to update the data in the matching relation database; and inputting the data in the matching relation database into the preliminary shoe tree data acquisition model for training to obtain a final shoe tree data acquisition model.

In the above solution, the step of obtaining the database of the matching relationship between the foot shape and the shoe last in terms of length, width and circumference index includes the following steps: obtaining the matching relation between the foot shape and the shoe tree in the length index according to the relation between the shoe tree bottom sample length and the foot length data, the back tolerance and the allowance; obtaining the matching relation between the foot shape and the width index of the shoe tree according to the relation between the width of the shoe tree, the foot width data and the allowance; obtaining the matching relation between the foot shape and the shoe tree in the circumference index according to the relation between the shoe tree circumference and the foot circumference data and the allowance; and storing the matching relations to obtain a matching relation database.

In the above solution, the establishing of the foot type data obtaining model includes the following steps: acquiring a plurality of foot images through web crawlers and camera shooting; making a data set by using the obtained foot images, and dividing the data set into a training set, a testing set and a verification set; inputting the training set into the established second convolutional neural network model for training to obtain a preliminary foot shape data acquisition model; and fine-tuning the preliminary foot type data acquisition model by a fine-tune method, modifying related parameters and acquiring a final foot type data acquisition model.

In the above solution, the step of creating a data set using the obtained foot images, and dividing the data set into a training set, a test set, and a verification set includes the steps of: marking the obtained foot image by a LabelImg image marking tool; scaling the marked foot images by a bilinear interpolation algorithm to enable the sizes of all the images to be preset values, and storing the scaled foot images; converting the foot image after the zooming processing from an RGB space to an HSV space through a brightness conversion formula, and turning over the foot image after the zooming processing; storing the foot images after brightness conversion and turnover processing; normalizing the stored foot images by a linear function normalization method to obtain a data set; the data set is divided into a training set, a testing set and a verification set according to a preset proportion.

In the above solution, the obtaining of the customized information and the foot shape measurement information of the customer includes the following steps: the customer sends shoe customization information to the shoe customization platform; the method comprises the following steps that a manufacturer obtains contact information of a customer and information of shoes designed by the customer independently on a shoe customization platform; acquiring video information of a customer foot through a camera; the customer sends the video information of the foot to the shoe customization platform.

In the above solution, the preprocessing the foot shape measurement information of the customer includes the following steps: acquiring a video file of a customer foot, and acquiring the total frame number of the video file; carrying out image correction processing on each frame image of the video through Hough transformation; starting from a first frame image of a video, acquiring an image difference value of a current frame image and a next frame image; and taking the current frame image with the image difference value larger than the preset threshold value as a key frame image for storage, and removing the current frame image with the image difference value larger than the preset threshold value.

In conclusion, the beneficial effects of the invention are as follows: customer carries out the shoes of independently designing through shoes customization platform, and send the video information of customization information and foot to shoes customization platform, shoes customization platform carries out the algorithm through the video information to the foot and handles the back and acquire customer's foot type data, again with customer's foot type data input to shoe tree data acquisition model and obtain the shoe tree data, the producer makes shoe tree and shoes according to shoe tree data, make the customer participate in the design process, and simultaneously, customer only need send the video information of foot, customer's convenience and satisfaction have been improved.

Drawings

The accompanying drawings, which are incorporated in and constitute a part of this application, illustrate embodiments of the invention and, together with the description, serve to explain the invention and not to limit the invention.

FIG. 1 is a diagram of the steps of a method of customizing a footwear of the present invention.

FIG. 2 is a diagram illustrating steps for building a three-dimensional model of a shoe according to the present invention.

Fig. 3 is a step diagram of importing a shoe three-dimensional model into a shoe customization platform for rendering according to the present invention.

FIG. 4 is a step diagram of rendering three-dimensional model data of a shoe by texture according to the present invention.

FIG. 5 is a step diagram of rendering three-dimensional model data of a shoe by mapping according to the present invention.

Fig. 6 is a diagram illustrating a procedure for constructing a last data obtaining model according to the present invention.

FIG. 7 is a step chart of obtaining a database of the matching relationship between the foot shape and the shoe tree in terms of length, width, and circumference.

FIG. 8 is a diagram of the steps for building a foot data acquisition model in accordance with the present invention.

Fig. 9 is a step diagram for creating a data set using the acquired foot images and dividing the data set according to the present invention.

Fig. 10 is a diagram illustrating steps for obtaining custom information and foot shape measurement information of a customer according to the present invention.

Fig. 11 is a diagram illustrating a step of preprocessing the foot shape measurement information of the customer according to the present invention.

Detailed Description

In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention will be described in further detail with reference to the following embodiments and accompanying drawings. The exemplary embodiments and descriptions of the present invention are provided to explain the present invention, but not to limit the present invention.

As shown in FIG. 1, a method of customizing footwear according to the present invention comprises:

step S1: building a shoe three-dimensional model, and guiding the shoe three-dimensional model into a shoe customization platform for rendering;

step S2: establishing a shoe tree data acquisition model, and importing the shoe tree data acquisition model to a shoe customization platform;

step S3: establishing a foot type data acquisition model, and importing the foot type data acquisition model into a shoe customization platform;

step S4: obtaining customized information and foot shape measurement information of a customer;

step S5: preprocessing the foot shape measurement information of the customer, and inputting the foot shape measurement information into a foot shape data acquisition model to obtain foot shape data of the customer;

step S6: inputting the foot shape data of the customer into a shoe tree data acquisition model to obtain shoe tree data;

step S7: and manufacturing a shoe tree according to the shoe tree data, and manufacturing shoes according to the manufactured shoe tree.

As shown in fig. 2, the building of the three-dimensional model of the shoe comprises the following steps:

step S111: establishing a contour part model of the shoe through Maya three-dimensional software;

step S112: and (4) carrying out detail adjustment on the established shoe by a subdivision modeling method to obtain a complete shoe three-dimensional model.

In this embodiment, the contoured portion of the shoe comprises the sole, the upper and the laces of the shoe, the sole comprising an insole and an outsole.

In this embodiment, the detail adjustment of the built shoe through the subdivision modeling method includes smoothing the upper.

As shown in fig. 3, the importing the shoe three-dimensional model into the shoe customization platform for rendering includes:

step S121: rendering the three-dimensional shoe model data through textures;

step S122: rendering the three-dimensional shoe model data through the map.

As shown in fig. 4, the rendering the three-dimensional shoe model data by texture includes the following steps:

step S1211: storing the three-dimensional shoe model data and rendering texture resources;

step S1212: carrying out coordinate transformation processing on the vertexes of the shoe three-dimensional model through a vertex shader, and carrying out cutting processing on the shoe three-dimensional model subjected to the coordinate transformation processing to obtain primitive coordinates of the shoe three-dimensional model in a cutting space;

step S1213: converting each primitive coordinate of the shoe three-dimensional model in the cutting space to a screen coordinate system, and performing rasterization processing on the primitive in the screen coordinate system to obtain a fragment;

step S1214: texture sampling is carried out through a fragment shader to obtain texture coordinates of each fragment, and the color of each fragment is calculated through the texture coordinates of each fragment;

step S1215: carrying out template test and depth test on each fragment, carrying out fusion processing on the color of the template test and depth test fragment and the stored color, and storing the fragment subjected to the color fusion processing;

step S1216: and acquiring the three-dimensional model of the shoe after texture rendering.

In this embodiment, the performing, by the vertex shader, the coordinate transformation processing on the vertex of the three-dimensional model of the shoe includes converting the vertex coordinates of the three-dimensional model from the model space to the homogeneous clipping space to obtain homogeneous coordinates of the vertex, and obtaining the normalized device coordinates of the vertex after processing the homogeneous coordinates by the perspective division method.

In this embodiment, in the template testing process, the shoe customizing platform reads the template value of the screen coordinate position of each fragment, compares the template value of the screen coordinate position of each fragment with the template reference value, and discards the fragments of which the template value of the screen coordinate position is less than or equal to the depth reference value of the template reference value; in the depth testing process, the shoe customizing platform reads the depth value of each fragment, compares the depth value of each fragment with the depth reference value, and discards the fragments of which the depth values are less than or equal to the depth reference value of the depth reference value.

As shown in fig. 5, the rendering the three-dimensional shoe model data by mapping includes the following steps:

step S1221: storing the rendering map resources, and storing and packaging the shoe three-dimensional model;

step S1222: mapping the rendering map and each part in the three-dimensional model of the shoe;

step S1223: and obtaining a three-dimensional model of the shoe after the mapping rendering.

As shown in fig. 6, the building of the last data obtaining model includes the steps of:

step S21: acquiring a matching relation database of the foot shape and the indexes of the shoe tree in length, width and circumference;

step S22: constructing a first convolution neural network model, inputting data in the matching relation database into the first convolution neural network model for training to obtain a preliminary shoe tree data acquisition model;

step S23: acquiring foot shape sample data of a tester and comfort information of the tester trying on a shoe tree corresponding to the foot shape;

step S24: adjusting the shoe tree corresponding to the foot shape of the tester according to the comfort degree information until the tester feels comfortable;

step S25: inputting the foot type sample data of the tester and the corresponding shoe tree data into a matching relation database to update the data in the matching relation database;

step S26: and inputting the data in the matching relation database into the preliminary shoe tree data acquisition model for training to obtain a final shoe tree data acquisition model.

In this embodiment, the obtaining of the foot type sample data of the tester comprises obtaining foot type point cloud data of the tester by a three-dimensional foot scanner, and obtaining foot length data, foot width data, foot circumference data and foot height data of the tester by processing the foot type point cloud data, wherein the foot length data comprises a foot length, a first metatarsophalangeal joint length, a fifth metatarsophalangeal end point length and a heel center position length, the foot width data comprises a basic width, a first metatarsophalangeal lining width, a heel center width and a metatarsophalangeal diagonal width, the foot circumference data comprises a metatarsophalangeal circumference, a anterior tarsal circumference and a pocket circumference, and the foot height data comprises a thumb height, a first metatarsophalangeal height, an anterior tarsal height, an lateral malleolus lower edge height and an arch height; the processing of the foot-shaped point cloud data comprises the steps of processing the foot-shaped point cloud data through an ICP algorithm to obtain a three-dimensional foot shape, aligning the bottom of the three-dimensional foot shape with a coordinate system, projecting the three-dimensional foot shape on the bottom surface, taking the middle of a middle toe to the middle of a heel as a central axis, taking the axial direction as an x-axis direction, taking the basic width direction of the foot as a y-axis direction, taking the direction vertical to the x-axis and the y-axis as a z-axis direction, taking the farthest span of the foot shape on the x-axis as the foot length, taking the farthest span of the foot shape on the y-axis as the basic width of the foot, obtaining the lower edge height of a thumb, the height of a first metatarsophalangeal toe, the height of a front tarsal bone, the height of an outer ankle and the height from the data of the foot shape on the z-axis, projecting the foot shape on an xoy plane to obtain the circumference data of the foot shape, finding characteristic points according to the outline graph projected on the plane, and constructing an auxiliary plane, the foot circumference data can be obtained by converting and calculating the length approximation calculation of the point cloud and the intersecting contour line of the plane.

As shown in fig. 7, the step of obtaining the database of the matching relationship between the foot shape and the shoe last in the length, width and circumference indexes comprises the following steps:

step S211: obtaining the matching relation between the foot shape and the shoe tree in the length index according to the relation between the shoe tree bottom sample length and the foot length data, the back tolerance and the allowance;

step S212: obtaining the matching relation between the foot shape and the width index of the shoe tree according to the relation between the width of the shoe tree, the foot width data and the allowance;

step S213: obtaining the matching relation between the foot shape and the shoe tree in the circumference index according to the relation between the shoe tree circumference and the foot circumference data and the allowance;

step S214: and storing the matching relations to obtain a matching relation database.

In this embodiment, in order to ensure the moving allowance of the foot in the shoe cavity and prevent the shoe from pushing against the foot, the shoe last base length needs to be increased with allowance based on the foot length, and meanwhile, the heel is of an arc structure, so that the shoe does not fall off when walking, and the shoe last heel should have a proper rear allowance.

In this embodiment, the obtaining of the matching relationship between the foot shape and the length index of the shoe last according to the relationship between the shoe last bottom sample length and the foot length data, and the back tolerance and the allowance includes obtaining the matching relationship between the foot shape and the length index of the shoe last according to the shoe last bottom sample length, namely the foot length + the allowance-back tolerance, where the allowance and the back tolerance are fixed values; the step of obtaining the matching relation between the foot shape and the shoe tree in the circumference index according to the relation between the shoe tree circumference, the foot circumference data and the allowance comprises the step of obtaining the matching relation between the foot shape and the shoe tree in the circumference index according to the shoe tree circumference-foot circumference-circumference difference, wherein the circumference difference comprises a metatarsus circumference difference, a prepositive tarsal bone circumference difference and a pocket root circumference difference, and the circumference difference is used for representing the value of deformation of the foot in the shoe cavity due to the movement of the foot.

As shown in fig. 8, the building of the foot-type data acquisition model includes the following steps:

step S31: acquiring a plurality of foot images through web crawlers and camera shooting;

step S32: making a data set by using the obtained foot images, and dividing the data set into a training set, a testing set and a verification set;

step S33: inputting the training set into the established second convolutional neural network model for training to obtain a preliminary foot shape data acquisition model;

step S34: and fine-tuning the preliminary foot type data acquisition model by a fine-tune method, modifying related parameters and acquiring a final foot type data acquisition model.

As shown in fig. 9, the step of using the acquired foot images to produce a data set and dividing the data set into a training set, a testing set and a verification set comprises the following steps:

step S321: marking the obtained foot image by a LabelImg image marking tool;

step S322: scaling the marked foot images by a bilinear interpolation algorithm to enable the sizes of all the images to be preset values, and storing the scaled foot images;

step S323: converting the foot image after the zooming processing from an RGB space to an HSV space through a brightness conversion formula, and turning over the foot image after the zooming processing;

step S324: storing the foot images after brightness conversion and turnover processing;

step S325: normalizing the stored foot images by a linear function normalization method to obtain a data set;

step S326: the data set is divided into a training set, a testing set and a verification set according to a preset proportion.

As shown in fig. 10, the step of obtaining the customized information and the foot shape measurement information of the customer includes the following steps:

step S41: the customer sends shoe customization information to the shoe customization platform;

step S42: the method comprises the following steps that a manufacturer obtains contact information of a customer and information of shoes designed by the customer independently on a shoe customization platform;

step S43: acquiring video information of a customer foot through a camera;

step S44: the customer sends the video information of the foot to the shoe customization platform.

As shown in fig. 11, the preprocessing of the foot shape measurement information of the customer includes the following steps:

step S51: acquiring a video file of a customer foot, and acquiring the total frame number of the video file;

step S52: carrying out image correction processing on each frame image of the video through Hough transformation;

step S53: starting from a first frame image of a video, acquiring an image difference value of a current frame image and a next frame image;

step S54: and taking the current frame image with the image difference value larger than the preset threshold value as a key frame image for storage, and removing the current frame image with the image difference value larger than the preset threshold value.

In this embodiment, when a customer designs shoe information autonomously, the customer logs in a shoe customization platform and accesses an established three-dimensional shoe model, and simultaneously, selects shoe types including sports shoes, high-heeled shoes, leather shoes, sandals and the like by logging in an interface of the three-dimensional customized shoes in the shoe customization platform, selects textures and maps of uppers, selects styles of laces, textures and maps of tongues, textures and maps of uppers, textures and maps of insoles, textures and maps of outsoles after selecting the shoe types, acquires a three-dimensional display map of the designed shoes after selecting approximate shoe sizes, and the customer selects customization or re-design according to a satisfactory degree.

In this embodiment, the obtaining video information of the customer's foot by the camera includes obtaining video information including a left instep, a left sole, an inner side of the left foot, an outer side of the left foot, a left heel, a right instep, a right sole, an inner side of the right foot, an outer side of the right foot, and a right heel of the customer by the camera.

In this embodiment, in the process of performing image rectification processing on each frame of image of the video, performing image rectification processing on each frame of image of the video may improve the rate of the foot-type data acquisition model; and the current frame image with the aberration value larger than the preset threshold value is taken as a key frame image to be stored, the current frame image with the aberration value larger than the preset threshold value is removed, useless video frames can be reduced, and data input to the foot shape data acquisition model is reduced, so that the operation rate of the foot shape data acquisition model is improved.

The above description is only a preferred embodiment of the present invention, and is not intended to limit the present invention, and various modifications and changes may be made to the embodiment of the present invention by those skilled in the art. Any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.