Human motion recognition based on hidden Markov models

Hidden Markov Models are widely used in forecasting the unknown sequence based on observation on outside system. In this paper, they are applied in Human Motion Recognition. With the human's silhouettes, the paper mainly deals with how to get the models of regular actions and combine them with HMM to recognize the motions of motive people. As for the localization on gray images of silhouettes, an algorithm combining silhouette contrasting and centroid tracking is put forward. The results show that the new algorithm has better performance.