Abstract: This paper describes a new tracking algorithm for extended target moving in complex conditions. The moment invariants are chosen as the tracking feature since the size, shape and orientation of the extended target would change distinctly during the moving period. The moment invariants are calculated based on edge region of the local target, through which the computation quantity is reduced drastically. Utilizing these moment invariants, a new roughprecise stages correlation tracking algorithm is proposed, in which the multistage and subsample are adopted to reduce the matching time. In addition, the new creditability evaluation is given by the correlativity of the sequence images, according to that, the method of template correction is established. Experimental results indicate the valid