Abstract
In this thesis, we introduce a hand gesture recognition system to recognize continuous gesture in simple background. The system consists of three modules: feature extraction, hidden Markov model (HMM) training, and gesture recognition using the HMMs. First, we apply the motion information to extract the hand-shape and apply the scale and rotation-invariant Fourier descriptor to characterize hand figures. Then we combine Fourier descriptor and motion information of input image sequence as our feature vector. After having extracted the feature vector, we first train our system using HMM approach and then use the trained HMMs to recognize the input gesture. In training phase, we apply hidden Markov Model to describe the gestures properties (generating the initial state probability distribution, the state transition probability distribution and the observation probability distribution) for each gesture. To recognize gesture, the gesture to be recognized in separately scored against different HMMs. The model with the highest score is selected as the recognized gesture. Our system consists of 20 different hand gestures. The experimental results show that the average recognition rate is 88.5%.