Application of Directional Wavelets to Image Compression

A novel unitary directional wavelet methodology producing six complex-valued subbands per level of decomposition has been previously introduced [1,2] as a method of spatio-frequency decomposition for image fusion applications. Properties of the decomposition include unitarity, sparse computation, energy-packing efficiency, symmetry, and sub-pixel registration of all subbands at the same scale. The directional subband decomposition generatess six directionally oriented subbands and a subsampled image at each level of decomposition, as illustrated below.