Optimal bandsplitting algorithm using rearranged spatial tree for adaptive subband image coding

An adaptive bandsplitting tree representation using wavelet packet or adaptive subband was exploited by various effective image coding methods. Unlike the works in the literature, in this paper we propose a cost function that includes both rate and distortion considering the spatial dependency of cross-subband coefficients. The spatial tree for octave-decomposed wavelet coefficients is expanded for the arbitrarily bandsplitted subbands, and the spatial tree is defined using the rearranged coefficients. Finally, experimental results for some images show that it generally tends to be effective to octave-decompose the lower frequency components and to uniform-decompose the higher frequency components and that our coder gives a better coding efficiency compared with various coding method previously reported