Human Activity Recognition via 3-D joint angle features and Hidden Markov models

This paper presents a novel approach of Human Activity Recognition (HAR) using the joint angles of the human body in 3-D. From each pair of activity video images acquired by a stereo camera, the body joint angles are estimated by co-registering a 3-D body model to the stereo information: our approach uses no attached sensors on the human. The estimated joint angle features from the time-sequential activity video frames are then mapped into codewords to generate a sequence of discrete symbols for a Hidden Markov Model (HMM) of each activity. With these symbols, each activity HMM is trained and used for activity recognition. The performance of our joint angle-based HAR has been compared to that of the conventional binary silhouette-based HAR, producing significantly better results in the recognition rate: especially for those activities that are not discernible with the conventional approaches.