Novel Machine Learning for Hand Gesture Recognition Using Multiple View

Abstract

Different from the conventional communication method between users and machines, we use hand gesture to control the equipments. This paper presents hand gesture recognition applied human-computer interaction (HCI) system. It presents new method to automatic gesture area segmentation and orientation normalization of the gesture. It is not mandatory for the user to keep upright gestures in the regular position, the system segments and normalizes the gestures automatically. The method is an unsupervised nonlinear dimensionality reduction approach that utilizes the local linearity to discover the low dimensional manifold embedded in the high dimensional space. This suggests that the method may preserve the neighborhood configuration for the nonlinear structure of the multi-view hand shape data distribution. The experiment shows this method is very accurate. The gesture pointing accuracy of our system is measured by 80 times of pointing recognition test, the success rate above 90%.