A 3D Hand-drawn Gesture Input Device Using Fuzzy ARTMAP-based Recognizer

Abstract

In this paper, a novel input device based on 3D dynamic hand-drawn gestures is presented. It makes use of inertial sensor and pattern recognition technique. Fuzzy ARTMAP based recognizer is adopted to realize gesture recognition by using 3-axis acceleration signals directly instead of reproduced trajectories of gestures. The proposed method may relax motion constraints during inputting a gesture, which is more convenient for user. This prototype of input device has been implemented on a remote controller to manipulate TVs. The recognition rate of 20 gestures is higher than 97%. It clearly shows the effectiveness and feasibility of the proposed input device. As a result, it is a powerful, flexible interface for modern electronic products.

Keywords

• Gesture Recognition
• Neural Network
• feature extraction