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

Abstract The use of hand gesture provides an attractive alternative to cumbersome interface devices for human–computer interaction (HCI). Many hand gesture recognition methods using visual analysis have been proposed: syntactical analysis, neural networks, the hidden Markov model (HMM). In our research, an HMM is proposed for various types of hand gesture recognition. In the preprocessing stage, this approach consists of three different procedures for hand localization, hand tracking and gesture spotting. The hand location procedure 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 algorithm divides the trajectory into real and meaningless segments. To construct a feature database, this approach uses a combined and weighted location, angle and velocity feature codes, and employs a k-means clustering algorithm for the HMM codebook. In our experiments, 2400 trained gestures and 2400 untrained gestures are used for training and testing, respectively. Those experimental results demonstrate that the proposed approach yields a satisfactory and higher recognition rate for user images of different hand size, shape and skew angle.