A new series of biorthogonal wavelet filters for image compression

Abstract

A technique for deriving a new series of symmetric biorthogonal wavelets from a given symmetric regular filter is presented. The main idea is to find a symmetric complementary regular filter of the given filter which has the least square amplitude deviation from the ideal half band filter. Thus from a given regular symmetric filter, or from a known symmetric biorthogonal wavelet, a new series of symmetric biorthogonal wavelet filter banks with better frequency selectivity can easily be designed. Once the complementary filter is found out, another complementary filter of this complementary filter can again be obtained in the same way. Particularly, a new series of biorthogonal wavelet filters are derived from the standard Anotonini 9/7 biorthogonal wavelets. Applying these wavelet to filters to compression of the well known images of Lena, Barbara and Goldhill, etc., improved results in PSNR, are achieved.