Reversible Wavelets and their Application to Image Coding

Abstract

In this paper, we propose one-dimentional reversible two-channel filter banks. Up to now, only reversible filter banks (reversible wavelet transforms) have been proposed which correspond to particular perfect reconstruction filters with (taps of lowpass filter)^*(taps of highpass filter) being 2^*2,1^*3,5^*3,2^*6. In this paper, we propose reversible filter banks which correspond to general perfect reconstruction filters with (2N)^*(2N), (2N+1)^*(2M+1). Firstly, reversible filter banks of 2^*2 and 1^*3 are shown and then various reversible filter banks are derived by combining them. The lossless and lossy compression efficiencies of the proposed method are compared with conventional methods in consideration of their application to unified lossless and lossy coding of still images. It is shown that compression efficiency of the proposed method is almost the same as conventional methods.