Natural Human-computer Interaction Techniques And System Realization

Natural human-computer interactive research is an important branch of human-computer interactive techniques. It mainly does some research about how to interact with computer in a natural way and it can be divided into voice interaction, gesture interaction and so on by the way you choose to interact. Compared with traditional ways of interaction, such as mouse interaction, button interaction and so on, natural human-computer interaction is no doubt more concerned by people. Nowadays, there are some improvements of natural human-computer interactive research. However, there is still few system using this technology. So, it is now a core problem about how to apply natural human-computer techniques to the application systems.This paper design and realize a natural human-computer system, which can be used in teaching, with the research of voice recognition techniques, pattern tracking techniques and gesture recognition techniques. We analyze the engine of Microsoft speech SDK and use it to do voice recognition. We design and realize pattern tracking and real-time recognition algorithms, which can make our interactions more effective. We improve gesture recognition algorithms, which make our recognition faster and more accurate.This paper mainly do research in:.(1) Design a human-computer interactive system. We use COM to divide interactive engine and UI module of our system, which can reduce our system's coupling.(2) Do research in laser point interactive techniques and design video based pattern tracking and recognition algorithms to realize laser pointer interaction parts, which allow user to do interaction with our system with laser pointer. (3) Do research in gesture interactive techniques and improve histogram based gesture recognition algorithms to realize gesture interaction parts, which can make the system to recognize your gestures.(4) Do research in voice interactive techniques to realize voice interaction parts, which can make the system to understand what you say.Our test results show that our system can do real-time interaction with laser pointers and gestures. Also, our system can recognize the voice of different people. And the time delay of our system is acceptable.