A Sound-based Gesture Recognition Technology Designed for Mobile Platform

Abstract

The paper presents a gesture recognition technology applied in mobile platform. Gesture recognition technologies have been widely used in human-computer interaction fleld, however, most vision-based technologies are generally sensitive to environment and equipment (e.g. limited by lighting conditions and sensor precision). Besides, the techniques hardly apply to the mobile platform because of the high computational complexity. We propose a sound-based gesture recognition technology supporting 5 recognizable gestures in mobile platform. The Goertzel algorithm is leveraged to reduce the computational complexity, and the frequency band-division technique is ofiered to improve the accuracy and the robustness of the recognition technology, what’s more, a semiautomatic threshold selection design will be provided to meet the difierent requirements of users. The paper introduces correlation theory and the implementation details, the robust experiments and results analysis are discussed in the end.