A nonlinear filter bank for image coding

A new nonexpansive pyramidal decomposition is proposed. The signal is decomposed through a nonlinear filter bank into low- and high-pass signals and the recursion of the filter bank over the low-pass signal generates a pyramid resembling that of the octave wavelet transform. The transformed samples were grouped into square blocks and used to replace the DCT in the JPEG coder. The proposed coder shows several advantages: computation is greatly reduced compared to the DCT, image edges are better encoded, blocking is eliminated, and it allows lossless coding.