Human Pose Detection Through Searching in 3D Database With 2D Extracted Skeletons

Abstract

With the proliferation of mobile devices, the world is saturated with data captured from these multi-sensor devices. Motion capture data is most commonly used for human figure animation. These data not only help us to create more realistic animation, but they also have been used for artistic purposes. With these data, we can model 3D animations that correspond to a human model. In this paper, we present a new way of comparing images of a human model in a stance to a database of 3D human figure poses. The results in this paper has shown that the new approach is both fast and accurate.