Tracking and identifying a changing appearance target

Abstract

Camshift algorithm has been popularly applied to tasks such as pedestrian tracking and traffic tracking. Camshift employs a histogram-based technique. The common hue-based histogram is robust to minor changes in the shape of a tracked target but it is only suitable to track a target having, relatively, a constant appearance. In this paper, we investigate the application of the Camshift algorithm and the Speeded Up Robust Features (SURF) in tracking and identifying a target. The target is tracked using the Camshift algorithm and identified using SURF. By combining the two techniques we show that a changing appearance target can be tracked and identifed at the same time.