Robust object tracking for resource-limited hardware systems

Abstract

Resource-limited hardware systems often generate LFR (low frame rate) videos in many real-world robot vision applications. Most existing approaches treat LFR video tracking as an abrupt motion tracking problem. However, in LFR video tracking applications, LFR not only causes abrupt motions, and also large appearance changes of objects because the objects' poses and illumination may undergo large changes from one frame to the next. This adds extra difficulties to LFR video tracking. In this paper, we propose a robust and general tracking system for LFR videos. The tracking system consists of four major parts: dominant color-spatial based object representation, cross bin-ratio based similarity measure, annealed PSO (particle swarm optimization) based searching, integral image of model parameters. The first two parts are combined to provide a good solution to the appearance changes, and the abrupt motion is effectively captured by the annealed PSO based searching. Moreover, an integral image of model parameters is constructed, which provides a look-up table for evaluation, and this greatly reduces the computational load. Experimental results demonstrate that the proposed tracking system can effectively tackle the difficulties caused by LFR. © 2011 Springer-Verlag.