The improvement of an image compression approach using weber-fechner law

Abstract

The idea behind image compression is the reduction of the average number of bits per pixel needed for its representation. Indeed, our study of still image compression is based on the non-conservative compression method. This means that the reconstructed image, after a compression/decompression cycle, will be different from the original one. In fact, such a difference brings about a degradation of the original image. In the first part of still image compression, we suggest a chain composed of a discrete wavelet transform followed by neural-networks quantization and a binary encoder. In this paper, we improve the quality of compression by adding a pretreatment phase through the use of the principle of Weber-Fechner law which considers the human-eye sensitivity to luminance as a logarithmic function.