3D Reconstruction of Full Human Body Using a Single Low Cost RGB-D Sensor

Sorosh JahandidehMohammad SaadatsereshtAli Aghassi
The 11th Iranian and the first International Conference on Machine Vision and Image Processing (MVIP 2020)

Abstract

Three-dimensional scanners are popular tools for 3D human body modeling in recent decades. This paper propose a method for 3D human body modeling using a single Kinect sensor, because of their low cost and versatility. 3D modeling with a single Kinect faces some challenges such as quality reduction of raw data due to more sensor-object distance and higher body deformation during the longer scanning process. In order to overcome the first challenge, Kinect rotation motor for scanning in different angles was used. And to solve the instability of the user, non-rigid registration method was exploited. The experimental result shows our rapid 3D human body modeling system has a high capability comparing to other similar systems. This system has a lower cost (less than 200 dollars), capacity of scanning in near distance without additional equipment such as turntables, and a higher quality of the final 3D model so that the details such as wrinkles and hair style are recognizable. By this system, scanning and 3D modeling of a person with a height of 1.8 meters at a distance of 1.4 meters (about 80\% of person height) can be performed with average 3.7 mm point cloud density and 1.4 mm spatial noises.

Keywords: 3D human body modelling, Kinect sensor, Non-rigid registration,3D scanner