Research And Application Based On Hand Gesture Recognition System

With the development of human computer interaction, computer virtual reality technology uses the voice, gestures and so on to interact with player real-timely. It simulates the human visual, auditory, tactile and other sensory organs function to enable people to be immersed in the computer-generated virtual realm. Gesture is as a natural, intuitive man-machine communication. Vision-based gesture recognition is the key technology of the next generation of human-computer interaction technology. As the uncertainty of the gesture itself and the differences in time and space characteristics, the research has become a challenging multidisciplinary research topic. This paper is on the basis of the project of "a360-degree virtual interactive experience", and it provides gesture recognition as the study and analysis of the research status at home and abroad by using the monocular camera. By studying the complex context of gesture segmentation, gesture tracking, edge enhancement, feature extraction, extract the fingertips and the use of the Surf Feature operator to identify static gestures and on this basis to establish the gesture recognition system to identify the fingertips and static gestures driven applications. This system designs multiply interactive applications including the gestures to simulate mouse, fingertips detection and pictures and movies browsing to interact with the computer real-timely. By capturing the static images and video files as the test data, the system has reached a good recognition rate and reached the desired experiment.The contents of this paper are as follows:(1) This paper firstly plots the face position by haar characteristics, then analyses the color components in the YCbCr color model space. After getting the range of the face color, use the range of color components to segment the skin color. At last, use the motion detection to filter some similar color region parts.(2) Use the detection of the convex hull of the gesture contour, the corner point, the outside circum-circle diameter and perimeter and surf features to recognize the static gesture.(3) This paper uses the corner point detection, the convex hull and the hull defection detection to detect the fingertips. The basic algorithm is:firstly detect the corner points of the binary hand image, use the contour to detect the convex hull and hull defect and the geometric characteristics to detect the optimal corner point.The features of this paper are mainly reflected in the following two aspects:(1) Use the motion detection and skin color model both to segment the hand region, it use the initial threshold value by capturing and evaluating the range of skin color of the face firstly and use the motion detection to detect the possible hand’s position, then use the morphology to filter the noisy area.(2) Use the corner point detection, contour convex hull, center of gravity and the geometric characteristics to extract the fingers. Then fit the hand polygon features combined with the surf characteristics to recognize the gesture.