Multiwavelet-based Image Compression Using Human Visual System Model

Multiwavelet has some important properties such as orthogonality, symmetry, and short support, which make up for the shortcoming of scalar wavelet. In this paper, we consider the problem of improving the performance of multiwavelet-based image coders combining human visual syetem (HVS). By taking into account the imperfections inherent to the HVS, HVS weighting is designed to achieve higher compression rate and minimizing distortion due to compression. After multiwavelet transform, coefficients in different subbands are weighted directly by the band-average of the contrast sensitivity function (CSF) curve in the normalized spatial frequency domain, at last multiwavelet-based set partitioning in hierarchical trees (MSPIHT) algorithm is used to code forming embedded bit stream. Experimental results showed our proposed image compression algorithm can get better subject visual quality than the conventional MSPIHT algorithm at the same compression ratio.