Gesture Detection Of RGB-D Image-based

Gesture recognition is the hotspot of the human-computer interaction, and gesture detection is one of the key technologies for gesture recognition. The traditional gesture detection methods need not just to use skin detection, but also to locate hand, which lead to high computational complexity. And thanks to emergence of the depth sensor, a novel idea is presented for hand detection. The depth images obtained by the depth one contain the depth information of the object.The hand gesture can be rapidly detected just by depth images, but its detection accuracy is low under the complex background.Now most of the gesture detection algorithms utilize either RGB image or depth image. But it is seldom to use RGB and depth images together. This paper puts forward a detection method based on RGB-D. The basic idea is global optimal way. First, the image data information is extracted. Then, the RGB-D gesture image segmented is optimized using the convex function to identify gestures rapidly and accurately. Finally, the model is solved by the Split Bregman algorithm using the method of minimizing function and function constrain to find the optimal RGB-D segmentation. The experiments show that this method improves the segmentation accuracy comparing with traditional methods and can obtain a good result of the gesture segmentation and under the situation of self occlusion of the part hand area and a complex scene and this algorithm has excellent distinguish ability and robustness.