Complex multiple features tracking algorithm in motion capture

To track complex multiple features in video sequences is always a challenging problem; we present a feature-tracking algorithm integrating feature recognition and feature matching in this paper. According to feature attributes and relationship among estimated features, extracted features are classified as four types of features. Then different quantitative matching strategies are applied to track different kinds of features. To verify the tracks, a cross correlation test and predicted 3D model based test are used to test and remove outliers. The contributions and characters of each attribute are considered in our algorithm. The experimental results demonstrate the efficiency of the presented motion-tracking algorithm.