Tracking color objects in real time
Vladimir Kravtchenko · cIRcle (University of British Columbia) · 2009
The goal of our research is efficient tracking of color objects from a sequence of live images for use in real-time applications including surveillance, video conferencing and robot navigation. In this work we outline the results of our research. First we propose a novel, compact, look-up table color representation of a dielectric object that models the behavior of a color cluster in color space and yields real time performance in segmenting out color object pixels. This representation accounts for non-white illumination, shadows, highlights, variable viewing and camera operating conditions. We then propose a clustering method that uses density and spatial cues to cluster object pixels into separate objects. We also describe a method of identifying objects from the neighboring frames and predicting their future movement. Finally we provide details of a practical implementation of a tracking system based on the proposed techniques.