Foot plant detection for motion capture data by curve saliency

Foot plant detection is a fundamental step of most motion editing researches. Due to the footskate of captured motions, it is difficult to precisely extract foot plants. In this paper, we present a general, efficient approach for foot plant detection. The key idea is determining the boundaries of foot plants by searching the most salient points around the boundaries. By this approach, foot plants can be precisely extracted from most of the flat terrain motions. The experimental results show the generality and robustness of our approach.