チラズ ベルアブデカデール, 高橋和彦, 大谷淳
Predictive Filtering for Robust 2D Feature Tracking in Real-time
Abstract:Recent years have seen increased interest in human motion analysis, due to its importance in various real-world applications, such as in advanced man-machine interface systems, visual communications (e.g. tele-conferencing) and virtual reality. Real-time 3D pose estimation of humans in video is a basic part' of many of these applications. The estimated pose data can be used to drive (animate) avatars in a visual communication environment, for example.