Dynamic Human Body Size Measurement Based on Feature Points Prediction and Mapping
Authors: Tan, X., Lv, Z., Wang, K. and Yang, X.
Journal: Communications in Computer and Information Science
In this paper, we tackle the problem of measuring the sizes in different 3D human body motion models. We consider the problem of unexpected sizes when customers buy clothes in online shops because of past static dimensions measurement cannot meet dynamic requirements. The main contribution is a method that helps customers buy suitable clothes with a depth camera at home. Firstly, a random forest regression model is used to get the location of semantic feature points. Secondly, the tracking results are calculated with the help of a function map method which maps the semantic feature points between source model and motion sequence. Finally, we can calculate sizes of motion models with semantic feature points, with which can help evaluate whether a specific clothing is fit to body when doing actions. The proposed method accepts automatically predicting the points on 3D human models with scale transformation and even partial loss. A wide variety of experiments are conducted in which the method proved to achieve a significant result for anthropometric measurement in motion.