A hidden Markov model based dynamic hand gesture recognition system using OpenCV

In this paper we propose a novel and faster system for dynamic hand gesture recognition by using Intel's image processing library OpenCV. Many hand gesture recognition methods using visual analysis have been proposed: syntactical analysis, neural networks, the hidden Markov model (HMM). In our research, a HMM is proposed for hand gesture recognition. The whole system is divided into three stages detection and tracking, feature extraction and training and recognition. The first stage uses a more non-conventional approach of application of Lαβ colour space for hand detection. While the process of features extraction is the combination of Hu invariant moments and hand orientation. For the training, Baum-Welch algorithm using Left-Right Banded (LRB) topology is applied and recognition is achieved by Forward algorithm with an average recognition rate above 90% for isolated hand gestures. Because of the use of OpenCV's inbuilt functions, the system is easy to develop, its recognition rate is quite fast and so the system can be practically used for real-time applications.