Edge-classified adaptive image compression using wavelet transform

Abstract

The WT(Wavelet Transform) has become a very attractive technology in image compression area, because of its multiresolution image decomposition property. This thesis proposes a new image coding technique based on the combination of edge classification and the statistical energy distribution of high frequency subbands which are decomposed by the WT. Several high frequency subbands with lower energy distribution are eliminated from coding process for high compression ratio in the proposed compressiom method. A classification procedure is subsequently used to classify the other high frequency subband blocks with higher energy distribution into a finite number of classes according to edge classification criteria. The classified high frequency subband blocks are thresholed and are quantized in order to meet the predetermined bit rate using the classification map. The lowest frequency subband is encoded by DPCM(Differential Pulse Code Modulation). PSNR(Peak Signal-to-Noise Ratio) of the compressed image is computed for given bit rates in order to analyze the compression performance of the proposed scheme. The computed PSNR is also compared to those of JPEG(Joint Photographic Experts Group) base-linescheme and VQ(Vector Quantization) based compression scheme.