Hand gesture recognition using combined features of location, angle and velocity

The use of hand gesture provides an attractive alternative to cumbersome in terface devices for human-computer interaction (HCl). Many hand gesture rec ognition methods using visual analysis have been proposed: syntactical anal ysis, neural networks, the hidden Markov model (HMM). In our research, an H MM is proposed for various types of hand gesture recognition. In the prepro cessing stage, this approach consists of three different procedures for han d localization, hand tracking and gesture spotting. The hand location proce dure detects hand candidate regions on the basis of skin-color and motion. The hand tracking algorithm finds the centroids of the moving hand regions, connects them, and produces a hand trajectory. The gesture spotting algori thm divides the trajectory into real and meaningless segments. To construct a feature database, this approach uses a combined and weighted location, a ngle and velocity feature codes, and employs a k-means clustering algorithm for the HMM codebook. In our experiments, 2400 trained gestures and 2400 u ntrained gestures are used for training and testing, respectively. Those ex perimental results demonstrate that the proposed approach yields a satisfac tory and higher recognition rate for user images of different hand size. sh ape and skew angle. (C) 2001 Pattern Recognition Society. Published by Else vier Science Ltd. All rights reserved.