Dynamic hand gesture recognition

Recognition of dynamic gesture is one of the most challenging tasks in the computer vision. This paper deals with dynamic hand gesture (digits) recognition. The method consist of three fold novel contributions: firstly finding the flow of hand, secondly recognition technique of signs (Dynamic digits) and thirdly classification of gesture. We worked on 11 to 20 (digits) gestures. Spatiotemporal volume is not always sufficient to grasp the flow of hand movement. We proposed new method to find flow of hand for special signs. Features are extracted using proposed feature extraction algorithm and gestures are recognized. Proposed algorithm shows recognition efficiency of 94% for dynamic gestures. Time complexity of the given method is less as compared to some traditional methods.