Reconstruction and Analysis of Three-dimension Motion Based on the Detection of Depth Information
Tan Ai-d · Jiangxi kexue · 2013
We describe a novel system that automatically evaluates trainee's performances against a gold-standard performance and provides visual feedback to the trainee in a 3D virtual environment.The system consists of three components,namely motion capture,motion rebuilding,motion analysis.Motion capture: Our work takes advantage of Kinect,the latest motion sensor from Microsoft,to estimate the data of key nodes of the human skeleton,and to convert them into a protocol format in order to match the skeleton hierarchy of the human model. Motion rebuilding: We combine the motion capture data and the personalized human body model exported from MakeHuman,and reconstruct the motion as human skeletal animation. Motion analysis: Euclidean-distance-based method and Slice-Match method were introduced into our system and experimental results show that the latter outperforms the former. Integrated three above-mentioned techniques,the proposed computer assistant sport system can reconstruct three-dimension motion and give suggestions corresponding to the trainees' faults.