The Research Of Human-computer Interaction System Based On Gesture

For the continuous expanding of computer applications in various area, the bottle-neck effect of existing human-computer interaction channel has become more and more prominent. Nowadays, as the promoting of pervasive computing and people-oriented concept, natural HCI becomes an important research area of computer applications. Gesture is a common adopted and expressive way for communicating with eachother, therefor it aroused the concern of researchers. It is no doubt that introduce gesture into HCI will effectively broaden the HCI channel and greatly enhance the user experience.Take the gesture signal acquisition method as the classification standard, the cur-rent gesture-based humand-computer interaction research can be classified as four main categories, data glove based, machine vision based, SEMG based and inertia sensor based. Although these solutions have their unique advantages, the inertia sensor based solution has a highlights advantage while considering the human, machine and envi-ronment factors comprehensivly. This study uses accelerometers to collect the hand movement data. The trajectory of moving hand can be obtained by processing the raw acceleration data. A vector quantization method is adopted to extract the feature of the gesture. The method can greatly reduce the time and space complexity of follow-up treatments. HMM is used to recognize the gesture, there is a model corresponding each specific gesture. To partition each gesture embedded in the continuous input stream, threshold model is introduced as a divider. The model can not only segment the atom gestures, but also can find the best match of the target gesture model. For improving the identification accuracy, the acceleration signal is denoised before extracting the gesture feature. Here uses a modified Kalman filter.Following the resolved issue of gesture recognition, the thesis discusses the design of dual-hand HCI. Starting from the theory of dual-hand HCI, the issues of behavior basis, psychology basis and the timing characteristics of dual-hand interaction are dis-sused. Taking the dual-hand interaction theory as the basis, the parameters coordination of gesture library design is dissused.After the theoretical study, the proposed theory and method has been tested with a usecase. The usecase is a image viewer, it supports a set of predefined gesture to carry out the control command of loading, translation,zooming and rotating. This usecase preliminary validates the feasibility of gesture-based natural HCI.