| With the rapid development of computer technology, HCI (Human-Computer Interaction) is becoming an important part of our everyday life. HCI technologies become the focus of research, including face recognition, speech recognition and behavior recognition and so on. Gesture is an important means which people use it for expressing their feelings or transmitting information. Gesture recognition, which has become an indispensable part of human computer interaction technologies, gets more and more attention and wild range of applications, such as virtual reality, medical research, sign language learning and intelligent monitoring and so on. So the study of gesture recognition has far-reaching meaning and social value.In this paper, the gesture recognition technology is studied. And the HMM algorithm is improved based on the study and analysis on HMM and SVM applied in the pattern recognition. The main idea of this paper is improving the gesture recognition algorithm using information fusion on decision-level. While identifying the posture using SVM, the dynamic gesture track is recognized using the strong classification ability of SVM and the time-varying signal processing of HMM. Then we'll get the final recognition results by information fusion. To improve the recognition efficiency, the gesture template library is designed to index table, which also has strong scalability.The system is constructed under the environment of Visual C++ 6.0. And the software MATLAB is used to realize some simulation experiments. In the system, it achieves hand gesture segmentation based on skin color detection, the gesture contour feature extraction and trajectory tracking. And at last, I use comparison experiments to analysis the improved system's performance.The experiment is carried on four special dynamic hand gestures. And the results show that the improved dynamic hand gesture recognition system has a high recognition rate up to 94.5% and has a higher recognition efficiency. |
PDF Full Text Request