Studies On Tracking And Recognition Of Real Time Hand Gesture Based On Vision

Hand gesture is a natural and direct interaction method in human, as human computer interaction shift from computer centered to people centered, studies on hand gesture recognition has become a hot research spot. It provides users with natural human computer interaction methods, and has a theoretical meaning and practical value.Hand gesture is a natural,intuitive,easy to learn human computer interaction method. It can be classified into hand gesture based on data glove and hand gesture based on vision. Hand gesture recognition based on vision mainly depend on human to make various gesture to control computer. Because of diversity,ambiguity, differences in time and space, plus hand is a complex deformed object and uncertainty of vision, hand gesture recognition based on vision is a challenging research topic.This topic mainly research on hand gesture recognition of entertainment robot based on vision, for the three predefined hand gesture types, a kind of algorithm combined with various image processing technique to distinguish hand gesture has been proposed. In addition to these three hand gestures,the algorithm can also recognize one finger, three fingers, four fingers precisely, which has higher scalability,Firstly,this paper research on hand gesture in the static image, in the prep-processing stage,it use split hand gesture from background by using skin color extraction method., then with use of method of dilation, erosion, filling the black area surrounded by white, removing the white independent blocks to get template hand gesture. It extract the number of fingers of template hand gesture, and decide the specific hand gesture by the numbers of fingers at last.In the dynamic tracking and recognition process, using double buffer mechanism, one buffer to store the newly current data, the other as a copy of the current data. Tlirough set a timer,update the copy buffer before every time call the process algorithm to recognize the current hand gesture. Implementing the recognition algorithm on vs2005 IDE, the experiment result show that system has higher recognition rate, robustness and rapid response. The disadvantage is that it will produce interference during gesture segmentation in complex environments, which will affect recognition result.