Gesture Recognition Technology Based On Binocular Stereo Vision

One of the hottest topics in the field of human-computer interaction (HCI) is how tooperate computers in a more effective and natural way in order to reach better userexperience. Gesture-based HCI system is more acceptable for human users than other HCIsystems. Gestures are divided into static and dynamic gestures. Generally, static gesture isthe shape represented by hands; dynamic gesture is the trajectory of hands in space.By introducing stereo vision technology into gesture recognition system, gesturerecognition technology is not limited by the information of2-dimensional (2D) planeimage and depth information can be used for developing effective image analyzing andunderstanding algorithms. The thesis studies at gesture recognition method based onbinocular stereo vision which includes gesture segmenting, tracking and recognizing. Asegmenting method based on depth image and background modeling is proposed, it cantackle the problem of loss of depth information caused by motion blur. A new trackingmethod combining color information with depth information which could real-time trackdynamic gesture with high accuracy is introduced. A feature extraction and classificationmethod for gesture and its shape is proposed for correctly distinguishing among gesturesoperated by different users at different distances which makes gesture recognition systemmore robust and more practical. By using the methods above, develop a gesture-controlHCI demonstration system based on Stereo Vision.Experiment has proved that the proposed method has high accuracy for both dynamicand static gesture, and can meet the requirements of real-time. The test with IPTVapplication shows that the human-computer interaction system based on gesture controlhas good interactive effects and user experience.