2D-SOFM Vector Quantization for Image Compression Based on Inverse Difference Pyramidal Decomposition

Abstract

In this paper a new developed algorithm for compression of still images based on 2D-SOFM NN's in correspondence with the method of inverse difference pyramid (IDP) decomposition is represented. The new developed algorithm is well suited to be used in progressive image transmission (PIT). Advantage of the method relies on the learning process and adaptation capability of NN's to reduce the matrices computation complexity and the total number of pyramid levels required for PIT. In addition to, for image reconstruction no interpolation is needed any more, which improves the quality of the reconstructed image