Using HMM to sign language video retrieval

This paper presents a system called DCMR. Content-based video searching is a challenging field, and most research focus on the low level features such as color histogram, texture and etc. In this paper, we solve the searching problem by high level features used by hand language recognition. Firstly, we find the face in video frames that has complex background, and then we find the left hand and right hand in specific areas. By computing the hands´ length, position, velocity, acceleration, Fourier figure descriptor and etc, we generate the hands´ dynamic features. Consequently, we segment the video frames by motion features. As for each segment, we generate a HMM. When a clip of hand language inputs, we also get the feature serials, and then we compare the possibility of the input serials in each HMM. Experiment results on a large of hand language videos show that our searching system performs much better than existing methods on hand language video searching systems. Compared with the traditional methods, our system reduces the average searching time by half and the searching precision has doubled.