A Novel Moving Object Tracking Method Using ICA-R

Abstract

In this paper, the independent component analysis with reference (ICA-R) algorithm is proposed to track moving object. Taking the invariant moment of the detected object image as the reference signal, the moving object which shares the same characteristic as the reference can be extracted from the video frames through the ICA-R algorithm. Our algorithm can easily be extended to tackle the non-totally occlusion problem: the detected object image is first divided into either two or four parts and the unoccluded sub-parts can then be tracked through the ICA-R algorithm. As a result, the tracking of the whole moving object can be realized. The experimental results demonstrate the effectiveness of the proposed method.