Geometry-based Dynamic Hand Gesture Recognition
Duc-Hoang Vo, Huu-Hung Huynh, Jean Meunier · The University of Danang - Journal of Science and Technology · 2015
Hand gestures play an important role in communication in the hard-of-hearing community. They are used to convey information instead of words. Besides, a system which is developed to identify gestures can be also used for human-computer interaction. In this paper, we propose a vision-based approach for recognizing dynamic hand gestures. Our processing consists of three main stages: pre-processing, feature extraction and recognition. The first stage involves two sub-stages: segmentation which locates the hand and extracts the corresponding silhouette using color information; separation that removes the arm based on geometrical properties. Some characteristics which describe the hand posture are then extracted. Finally, the recognition is performed using two popular algorithms, which are Dynamic Time Warping and Hidden Markov Model. The experiment is conducted on SKIG dataset with a comparison of classification accuracies corresponding to the two mentioned methods.