Research On Key Technologies Of Human Body Shape Parameter Acquisition And 3D Reconstruction Based On Photos

With the rapid development of digital image processing and three-dimensional human body modeling technology,three-dimensional human body model has been widely used in virtual fitting,games,security and vehicle design.With the continuous progress of technology,people are increasingly demanding the individualization and reality of three-dimensional human body model.How to quickly and efficiently construct a three-dimensional human model close to the real human body has become a new technology hotspot.to construct a three-dimensional human model which is close to the real human body.The idea of this paper is to extract the human body from the photos and then calculate the body shape data by feature recognition,and then generate a three-dimensional human model close to the real human body.The first chapter summarizes the research background and significance of this topic,and introduces the research status and development trend of digital image processing,as well as the research status and development trend of three-dimensional human body modeling.Finally,it introduces the main research contents of this topic.The second chapter mainly studies and designs the whole scheme of digital image processing,including the reading of user photos,the pre-processing of digital images and the analysis of human contour extraction algorithm of digital images.After collecting the user's photo information,the user's photo is pre-processed.After comparing several thresholding algorithms,the adaptive thresholding segmentation algorithm which is more suitable for this topic is selected.Then the image is filtered to reduce the noise of the image,and the result is obtained by convex hull detection.The general outline of the human body.The human body is extracted from the background quickly and accurately by the improved image segmentation algorithm.Finally,the final human contour image is obtained by the open-close operation and edge detection of the image.The third chapter mainly studies and designs the acquisition scheme of human body shape data.Firstly,it defines the human body's characteristic points and the human body's baseline.Then,it identifies the human body's characteristic points and marks the characteristic points through the eight-chain code algorithm.Then,it calculates the human body's circumference size and curve size by two methods.Finally,it adopts the eight-chain code algorithm to identify the human body's characteristic points The binary first order model and polynomial fitting model are used to fit thecircumference size and curve size respectively,so as to obtain accurate human body feature size data.The fourth chapter mainly designs the scheme of three-dimensional human body modeling.Firstly,the suitable three-dimensional human body model is matched from the standard human body model library by human body feature size matching,feature point matching and contour matching.Then,through horizontal and vertical displacement transformation algorithm,special part transformation algorithm and radian transformation algorithm,the matched three-dimensional human body model is further optimized to make the generated three-dimensional human body model closer to the real human body.Finally,the human body model is rendered by local rendering method,which greatly improves the rendering speed and effectively reduces the computer load.The fifth chapter mainly introduces the development environment of this paper and the analysis of the experimental data generated in this development environment.The data analysis is mainly divided into image segmentation effect analysis,human body shape data fitting analysis and three-dimensional human body modeling effect analysis.The sixth chapter mainly summarizes and prospects the research and design of this topic.