A comparison of 3D hand gesture recognition using dynamic time warping

To enhance a hand gesture recognition system, we compare performance in accordance with various parameters. We present an efficient framework for gesture recognition that can be easily implemented with low computational costs. Based on the simple K-NN classifier, we develop a pattern matching method through combining the Dynamic Time Warping (DTW) alignment and distance measure for similarity between two sequences. In this process, we extract various features of hand and apply various distance measure for similarity. In addition to the gesture features and distance measures, we proposed preprocessing method to enhance the performance.