Research And Implementation Of Garment Customization System Based On Human Image Processing

In recent years,a new business model combining apparel customization with online e-commerce has quietly emerged as consumers’ demand for personalized clothing increases and e-commerce flourishes.This model aims to provide consumers with personalized,customized clothing to meet their preferences and needs.However,this model also faces two major challenges,namely the online fitting experience and remote body sizing.The online fitting experience refers to the process of simulating the actual fit of a customized garment on a user through a digital platform,and remote body measurements refers to the process of capturing images and obtaining the dimensional parameters of the user’s body parts through a digital device such as a smartphone camera.Virtual fitting and remote anthropometry are crucial to increase users’ confidence and satisfaction in purchasing customized clothing and to improve brand influence.To address these two challenges,the research in this thesis is as follows.1.Improvement of CP-VTON+ two-stage virtual fitting algorithm.Firstly,a perceptual loss function is introduced in the geometric matching module to enhance the retention of clothing features in complex scenarios.Secondly,in the virtual fitting module,a parallel hybrid attention mechanism is proposed and applied to U-shaped network.Meanwhile,residual blocks are used in U-shaped network layer to improve the stability of the model.The comparison experiments with other methods prove that the improved method proposed in this thesis not only improves the fusion effect of target body and target clothing,but also better preserves the features of other non-target clothing areas.2.Improvement of model-fitting-based anthropometric algorithms.The 3D reconstruction of human body image with orthogonal view is used to make up for the shortage of depth information in the single view image.At the same time,the method of dimensional measurement of various parts of human body based on 3D human body model is improved.By selecting marks on the body surface,the dimensional data is closer to that of manual measurement.3.The design and implementation of a garment customization system based on body image processing.Based on online virtual fitting and remote anthropometry,the system uses a variety of enterprise-level application development technologies to provide users with a convenient and efficient way to customize their garments.The clothing design module of the system allows users to choose different clothing colors and patterns according to their preferences;The virtual fitting module allows users to preview the effect of their designed clothes on the target human body online,increasing users’ participation and satisfaction in the process of clothing customization;The remote body measurement module allows users to obtain measurements of various parts of their body by uploading their own photos,making customized clothes fit better.