A better approach for object tracking using dual-tree complex wavelet transform

One of the most crucial tasks in content based video coding is the object retrieval and its tracking in subsequent frames. A number of algorithms for object tracking using real wavelet transforms have been proposed in literature. Major deficiencies in these algorithms are due to shift variance and directional selectivity problem of the real wavelet transforms. In this paper an algorithm for object tracking using dual-tree complex wavelet transform has been proposed which removes these problems of real wavelet transforms, experimental results show that the proposed algorithm is better than existing wavelet transform based.