Abstract:With the expanding application of virtual reality technology, the importance of humancomputer interaction system has become increasingly prominent. The key index in the humancomputer interaction system is the accuracy of gesture detection and tracking. Therefore, a gesture detection and tracking algorithm suitable for virtual reality humancomputer interaction system is proposed in this paper. Firstly, based on the gesture contour model the gesture detection is used to detect the gesture, which improves the robustness of gesture detection effectively. Then the state space probability prediction is used to implement the gesture tracking. In addition, a trained Bayesian classifier is used to classify the gesture images to be detected. The experimental results show that, as compared with the traditional algorithm, the proposed algorithm has high realtime, accuracy and robustness, and it can effectively identify the movement gesture to meet the requirements of humancomputer interaction.