Low-aliasing wavelets for pyramidal image coding

The selection of filter bank in wavelet compression is crucial, affecting image quality and system design. In this paper we address the design of wavelet filters suitable for high quality resolution-scalable (pyramidal &-progressive) subband coding of images. Recently, the biorthogonal coiflet (cooklet) family of wavelet filters has been constructed and explicit frequency domain formulae have been developed. In this paper we construct previously unpublished BC 52/60 and BC 28/36 wavelet filter pairs. Simulation with SPIHT demonstrates that the new filter pairs offer full resolution compressed image quality only 1 dB worse than that achieved by the popular CDF 9/7 filters, while reducing aliasing error in low pyramid levels (taken from low subbands) by 5 dB and 4 dB PSNR respectively. That is, full resolution image quality is imperceptibly diminished, while aliasing errors in low pyramid levels are visibly reduced.