Remote Body Measuring And Personalized Female Mannequin Modeling Based On Images

With the development and popularization of network and computer technology, more and more people choose to purchase clothes online. Besides, quick personalized made-to-measure has become possible. Fitness is the basic requirement when people choose garments. Compared with traditional clothing shopping, shopping clothing online is lack of experience impression, so how to phase fit clothes online has become major concern of most consumers. It is the method to solve this problem that apparel e-commerce platform recommend clothes to consumers according to their size and provide 3D virtual fitting service. However, these methods including customization are all related to acquisition of consumers’ body data. In order to ensure convenience and high efficiency of apparel e-commerce, acquisition of body size should be quick and convenient. However,traditional manual measurement apparently cannot satisfy the need. Currently, popular 3D body measurement instruments are bulky and expensive, so they are not suitable for widespread use. Therefore, it is very necessary to study a quick and convenient method to measure human data remotely and generate personalized virtual human body.The main purpose of this study is to construct a convenient and fast method which can be used to measure human body size remotely and research how to model personalized virtual 3D human bodies by adopting the measured data. Methods of remote human body size measurement based on 2D images and modeling personalized virtual human bodies are put forward in this research. The specific contents are as follows:(1) Human body contour extractionHSV color model is adopted to extract contours in the remote human body measurement system. A method is put forward to directly extract human body contour from color images without converting them into gray images. Threshold of each HSV component is determined by drawing and analyzing color histograms. Then, body contours are segmented from the background. The whole process is completed by a program code written by MATLAB. Ten subjects were selected randomly to be taken front and sideimages under different conditions. Their contours could all be extracted completely in the method, which verified that the body contour extraction method proposed in this paper is feasible and accurate.(2) Characteristic human body size extraction and validationAfter obtaining contour of human body contours, 2D body size of forty-six users was extracted in the size extraction method proposed by other members of our project group.Measurement parts are as follows: neck width, neck thickness, shoulder width, chest width,chest thickness, underchest width, underchest thickness, internipple distance, wide waist,waist thickness, abdomen width, abdomen thickness, hip width, hip thickness, thigh width,thigh thickness, knee width, knee thickness, calf width, calf thickness, arm length.Measurement results obtained by the method in this paper were compared with that obtained by the two-dimensional non-contact human body measurement system by conducting T-test. Results of T-test show that there are no significant difference in sixteen measurement parts, and there are significant difference in neck thickness, shoulder width,internipple distance, calf width arm length. That is to say, measurement results obtained in the two methods are roughly coherent, which shows that contour extraction method proposed in this research is feasible.(3) Personalized 3D female mannequin modelingYoung female mannequins in Poser library were taken as basic models. Python script language was used to write a program to change mannequins proportionally and the regression relationship between body circumference and coordinate points was established.Rules was set up for all points on mannequin surface in the process of parameterization.Finally, personalized virtual lifelike models with smooth surface were established.(4) Establishment of personalized 3D female mannequin modeling systemA program to model personalized mannequin automatically was written by using OpenGL as figure port and visual C# as developing platform. 3D female mannequin modeling system was established based on remote measurement, which can automatically read 2D human body feature sizes obtained by remote measurement system and can import mannequins in 3DS format. Then these mannequins imported are parameterized and presented in this system.Remote human body measuring was realized and personalized 3D female mannequin modeling system was established in this study, which lay a foundation for patternauto-generation system and 3D virtual fitting system. Besides, results in this study can provide reliable basis for the development of apparel e-commerce.