Shape-Based Human Activity Recognition Using Edit Distance

In this paper, we present a new posture classification system to analyze different human activities directly from video sequence. For well recognizing each posture of an activity, we propose an adaptation of Radon transform called R-transform, which is invariant to common geometrical transformations, to represent low-level features. The advantage of the R transform lies in its low computational complexity and its robustness to frame loss in video, disjoint silhouettes and holes in the shape. The nice ability of posture classification can help us generate a set of key postures for transferring an activity sequence to a set of symbols. Then, a novel string matching scheme based on edit distance is proposed to analyze different human activities. Our experiment results show that superior recognition is achieved with our proposed method.