Research And Application Of Hand Gesture Recognition Technology

Vision based hand gesture recognition is a vital research topic in human-computer interaction, which is used to generate a more natural interface between human users and computer-based applications, and the trend of it has broad prospects in computer-based applications. With the gradual increase of computer performance and the development of Virtual Reality in recent years, hand gesture recognition technology as the basis of Virtual Reality technology has made great progress.However, hands are complex deformed, and the hands are diversity and uncertainty with time and space, so the human-computer interaction based hands is still in its experimental stage currently, and it has not been used widely nowadays, and the research of hand gesture recognition is a interdisciplinary and challenging research.The process of vision-based hand gesture recognition can be divided into four stages which is image pre-processing, gesture segmentation, feature extraction and gesture recognition. When I studied in Hochschule Reutlingen University, I took part in the Multi Touch project, and did some research on finger gesture recognition, and I also read and summarized much English and Chinese articles about vision-based hand gesture recognition. Based on it,I did some work which is as following:(1)1 summarized the some kinds of methods about gesture recognition technology by reading a lot of articles in the field of gesture recognition. In gesture segmentation technologies, I summarized two kinds of gesture segmentation technologies based on skin color model. In feature extraction part, I researched two methods of feature extraction about geometric properties and topological properties separately. I also researched the recognition technologies in static gestures and dynamic gestures.(2)1 did experiments in some key technologies about gesture recognition, including smooth, sharpen in image pre-processing, and segmentation with skin color model in gesture segmentation. After the experiments, the conclusion is that the simpler the background is, the better the result is. That is to say, when the background is black, white or some other color, the hand gesture can be recognized well. (3)I designed and completed the system of Picture Viewer based Hand Control which is coded in platform VS2008 with OpenCV2.0. The software uses smooth and sharpen of image pre-processing methods. The gesture segmentation is skin color model, and the feature extraction about geometric properties is used here. At last, the hand gesture is recognized by BP Neural network method.(4)Gesture controlling is applied in Picture Viewer, the user can use hand to control the software directly without the restriction of input devices. Not only the users operate the software more convenient, but also hand controlling interface is in accord with the communication between people. The harmony human-computer interaction meets the trend in artificial intelligence.