Wavelet Lossy Image Coding with Edge and Texture Preserving Using a Modified SPIHT

In this paper we address the problem of design a wavelet lossy image coder that allows edge and texture preserving using a modified version of SPIHT coder. First, we obtain the edge map of an input image by means of the SUSAN edge detector. Second, a reversible wavelet transform is made. Third, the coordinates of edges or textures are mapped to the correspondent position in the wavelet domain. Finally, the coding stage is made with SPIHT by changing the concept of significance, in which a coefficient is selected not only by magnitude but as part of feature of interest map, then the bit rate spend more bits in that zones and the bit stream is entropy coded. We show a set of experiments in order to demonstrate the effectiveness of the coder even at very low bit rates