Research And Implementation Of Gesture Recognition System

With the continuous development of computer science and technology, human-computer interaction technology has achieved more and more attention, and how toimprove user’s feeling of the production, for example, comfort and naturalness, is animportant research direction of many manufacturers today. On the PC, mobile phones,game consoles and other areas, people has produced many excellent interactive productswhich comply with our people’s habit, such as: iphone, kinect and so on. But in thetraditional television and interactive smart TVs, we still lack of a further study. Gesturerecognition technology is a key technology in human-computer interaction, how to usethis technology to manipulate the smart TV will be a very attractive research topic.In this situation, gesture recognition system were studied and analyzed in this article.Based on the existing static gesture recognition methods, we design the processes of thissystem. This system is divided into the following modules: Gesture image acquisitionmodule, the preprocessing modules, edge detection and image feature extraction module,and gesture recognition module. In image acquisition stage, we get gesture image inputthrough the USB camera. In preprocessing part, we compare the adaptive thresholdsegmentation with the overall fixed threshold segmentation. Then, after pretreatment, usecanny edge detection algorithm on the binary image, and draw out the outline map. Thenext step is extract the seven parameters in the outline map which based on geometricmoment invariants, formed a feature vector. Finally we select the template matchingmethod which based on the Euclidean distance for recognition.In the actual system testing, we choose five kinds of example gesture to establish asample template library, and then,take a identification of the sample in two differentbackground conditions. Real-time and system recognition rate reached a relatively goodlevel and reached the system’target.