Dynamic Gesture Recognition Research

The research of Human Computer Interaction (HCI) is a hot and important issue from the time of computer invention.Intelligent HCI is mainly based on voice, facial expressions, body language and gesture to interact between human and computer. Due to more consistent in the natural characteristics to human, thus the interactive experience of intelligent HCI can not be achieved in traditional interact way. Thus in this article, we will research the dynamic gesture recognition algorithm, in order to achieve better HCI effect.In this paper, the main problem to be solved is based on the multiple camera wide Angle under the condition of orthoscopic splicing detection tracking and gesture recognition. Gesture recognition is a challenging field of computer vision research.Because of the complexity of background, appearance variation, illumination changes and occlusions, visual tracking has many problems to solve.we propose a novel tracking framework that adapts various appearance changes of object and also owns the ability of reacquisition after drift.For the detection of hand, we selected the most popular descriptor HOG feature as our sample characteristics. HOG feature is widely used for a variety of target detection, target detection of pedestrians, vehicles, faces and even seabirds, a large number of experiments show that HOG is a highly descriptive operator. In order to achieve real-time requirement, a variety of optimization measures is utilized here, include integral image and skin segmentation etc.We also present a refinement strategy to improve the tracker’s performance by discarding the support vector corresponding to possible wrong updates by a matching template after reinitialization. Algorithm of the experimental results show that the algorithm is compared with other algorithms have the advantage on the test data set.