Design And Development Of A Garment And Human Body Measurement System Based On Photography

With the rapid development of Internet,the online retail sales have been the universe trend for the garment industry and the online shopping platform has also transferred from PC to mobile.Nowadays,the high return rates seem to be the biggest problem for global retailers.Research shows that the basic reason for most of the returns and exchanges is due to the inappropriate size.The lack of uniform industry standards in clothing measurement and the inadequate knowledge of the body size for customers are the key factors leading to the size problem.In this paper,a method of measuring clothing and human body size based on photography is proposed.The purpose is to design convenient and reliable tools for users to understand the matching degree between body and clothing parameters.It not only provides the garment measurement technology for garment manufacturers but also provides users with mobile-end anthropometric tools,to recommend the most appropriate size of product to the customer through automatic comparison.Firstly,a garment measurement system is designed,including a series of functions such as image capture(hardware),garment type classification and size measurement(software).In this thesis,the chessboard calibration method is applied to eliminate lens distortion and image preprocessing technology is used to extract contours.Moreover,garment templates are introduced to recognize garment types and feature points,which can identify 24 garment subcategories and numbers of key sizes efficiently.Secondly,an anthropometric system is designed based on Android terminal.It preprocesses the frontal and lateral photos taken under the simple daily background,and extracts the contours of the human body by improved image processing methods such as Canny,Hough transform and morphology etc.Meanwhile,it defines several scanning tools to retrieve the key feature points hierarchically with the twice-location strategy.Furthermore,a robust validation-checking mechanism is proposed based on rules.Then,the key human dimensions are calculated by regression formula.Finally,this dissertation presents a size recommendation method utilizing the clothing and body measurement results.It establishes a fitness-assessing model by analyzing human body shape and clothing profile,and recommends appropriate size for customers.Experiments show that both clothing and human body measurement methods have good performances in time efficiency and accuracy.And it is proved that the proposed clothing size recommendation method can provide practical value for online shopping.