Video Tracking Using Acoustic Triangulation

Abstract

This study focuses on the detection and triangulation of sound sources. Specifically, we focus on the detection of sound in order to track a person’s position with a video camera. Acoustic tracking, an alternative to visual tracking, is relatively inexpensive, passive (does not emit energy), and effective in low lighting environments [3]. Our project is broken into two major aspects: accurately discerning input as opposed to background noise and the localization of the sound source. In order to focus on the input signal, we analyze two methods: time averaging and impulse culling. After the sound is analyzed and filtered we focus on the triangulation of the source in 2-D space using direct and estimation techniques requiring three microphones. This process is geared towards finding a compromise between performance and complexity which allows implementation on a standard micro-controller.