Realtime user interface using particle filter with integral histogram

In order to apply a vision technique to the real-time user interface, a fast and effective method of tracking human face and hands, and recognizing human pose by inverse kinematics is suggested in this paper. The suggested method shows that when observation likelihood is estimated, it is able to process in high speed proportional to numbers of samples with the integral histogram. The experiment results show that the method using integral histogram is faster than the method based on existing color histogram, while keeping similar recognition performance. The proposed user interface system is ideally suitable for gaming devices that utilizes user gesture and pose as control input in real-time.