New image compression method using logarithmic quantization

Abstract

The compression methods of images based on neural network and wavelet transform gives a limitation in the compression ratio and the reconstructed image quality. To improve the compression ratio without quality degradation of the reconstructed image, we add a phase of pretreatment before the discrete wavelet transform uses the principle of Weber-Fechner law which says that the sensation of the human eye to light is logarithmic. We quantify the image by using the Weber-Fechner law to reduce the entropy of the image before the compression.