Novel 3D wavelet decomposition approaches for image compression

This paper describes three different approaches suitable for the implementation of the 3D Haar Wavelet Transform (HWT). These transforms are important in 3D image compression and video processing applications. The algorithms for the Standard (SHWT), Non-Standard (NSHWT) and Tensor Product (TPHWT) based decompositions are described together with their implementation in MATLAB(R). The influence of both the hard and soft threshold applied to these decompositions is also examined. Extensive experiments have been carried out on 3D medical images and these have shown that the non-standard approach exhibits the best performance in terms of computation time, the number of zero coefficients achieved and the quality of the reconstructed image.