Hand Gesture Recognition Using Haar-Like Features and a Stochastic Context-Free Grammar

This paper proposes a new approach to solve the problem of real-time vision-based hand gesture recognition with the combination of statistical and syntactic analyses. The fundamental idea is to divide the recognition problem into two levels according to the hierarchical property of hand gestures. The lower level of the approach implements the posture detection with a statistical method based on Haar-like features and the AdaBoost learning algorithm. With this method, a group of hand postures can be detected in real time with high recognition accuracy. The higher level of the approach implements the hand gesture recognition using the syntactic analysis based on a stochastic context-free grammar. The postures that are detected by the lower level are converted into a sequence of terminal strings according to the grammar. Based on the probability that is associated with each production rule, given an input string, the corresponding gesture can be identified by looking for the production rule that has the highest probability of generating the input string.