Vision-based 3D Shape Reconstructure of a Non-Rigid Object
Zhu Fang-wen, Gong Zhenbang, Ze Qin Cui · 2006
A method for positioning the feature points on a nonrigid object that have small nonrigid motion are described. It is applied to reconstruct a human being's hand gesture in 3D space in a vision-based mechanical gripper control system. The positioning method provided a simple way for matching feature points between images in the stereo vision system. The result is used on a real time control system of a dexterous gripper and the gesture information is converted to the action gesture of a dexterous gripper by a specially designed algorithm, it also can be used to reconstruct a virtual hand skeleton