Image-based Human Body Size Measurement Method And System Desig

Body size is the key reference data for measuring human body shape,and body size measurement is also a key technology for the development of e-commerce and clothing customization.With the development of the online shopping industry,whether online shopping clothes fit the body and the customization of clothing are inseparable from the accurate measurement of human body size.The body size measurement system can provide reliable body size data according to different users.Traditional body size measurement methods cannot achieve remote and convenient measurement,so image-based remote body size measurement has become a current research hotspot.This paper studies and designs a set of image-based remote human body size measurement methods and systems.By optimizing the u-net neural network,a human body contour extraction method is proposed,and the human body feature points are extracted by combining the lightweight openpose human body pose estimation algorithm and the human body rectangular frame.The height ratio method and the curve fitting method complete the body size measurement algorithm.By developing a body size measurement We Chat applet,the design of a remote and convenient body size measurement system based on images is realized.The main research works are as follows:1.Research and compare edge detection algorithms,analyze first-order and second-order edge detection operators,and neural network edge detection algorithms,and discuss the principles,advantages and disadvantages of human contour extraction methods.Design the u-net human body contour extraction network suitable for We Chat applet,deepen the network depth to obtain finer contour details,design the residual module and bilinear interpolation algorithm,use the canny operator to extract the contour and design the human body rectangular frame,Experiments show that the method in this paper can effectively improve the accuracy of human contour extraction and reduce the amount of calculation.2.By designing a lightweight openpose human pose estimation algorithm,a human body feature point extraction algorithm is proposed.Change the openpose algorithm VGG-19 to Mobile Net V1,obtain a model with less computation and faster-running speed,quickly and accurately obtain 18 key points of the human body,and the algorithm analyzes the accuracy of each key point to obtain the best solution for refining and segmenting the human body contour,After a series of experiments,the human image is divided into gender and position to determine the feature point coefficient,and the feature point extraction is realized.3.Combined with the extraction results of human body contours and feature points,according to the measurement rules and size fitting methods,the human body size is divided into two-dimensional size and three-dimensional size.The clothing size is recommended and the standard weight and body shape tips are given,and the reliability and stability of the fitting size are verified by experiments.4.In-depth analysis of the advantages and disadvantages of the We Chat applet,mobile APP,the We Chat official account and H5 convenient application,research on the design and development of the We Chat applet,propose a camera calibration method,design user information upload,shooting guidance and result from display Interface,build a database and cloud server to call and store user information and body size data,and complete the design of the body size measurement system.Experiments show that the average error of the system is about 2.5cm,and the overall measurement time is about 2 seconds.After practical application,more than 500 users of the system have completed online clothing customization.In summary,the body size measurement system designed in this paper can realize body size measurement and provide users with reliable body size reference data,which has a certain role in promoting the development of the electronic clothing customization industry.