Particle filter based moving object tracking with adaptive observation model

Abstract

In this research, we've tried to develop a method with background subtraction, distance measurement and color histogram with particle filter, to track any single moving object. In visual moving object tracking, the appearance of both objects and the surrounding scenes may experience enormous variations due to changes in the scale and viewing angles, or partial occlusions. Also the objects and the backgrounds may have confusing color. These challenges may weaken the effectiveness of a dedicated target observation model when based on color feature. Background subtraction helps, to eliminate unnecessary regions, to track even when the target object and the background has similar color and thereby reduces the number of particles as well as the execution time and cost. Moreover we use distance measurement information, to make the tracker successful, when there are several objects with similar color. Experimental results have been presented to show the effectiveness of our proposed system.