WAVELET COEFFICIENT QUANTIZATION METHOD APPLYING HUMAN -EYE VISUAL MODEL IN IMAGE COMPRESSION

摘要

The present invention relates to a wavelet coefficient quantization method applying a human-eye visual model in image compression process and especially adapts to remote sensing image compression. The present invention is based on human-eye visual characteristics. A wavelet domain visual quantization model is obtained on the basis of experiments in order to reflect the relationship between block distortion of wavelet coefficients and human-eye visual characteristics. The model is comprised of luminance components, masking components and frequency components. Wherein the luminance components are acquired from low-frequency coefficients after the k^th level wavelet transform; the masking components are acquired from the high- frequency coefficients of the second level and the third level; and the frequency components are acquired by statistical method. The method can be combined with each of the mainly used wavelet compression methods such as Embedded Zero-tree Wavelet (EZW), Set Partitioning In Hierarchical Trees (SPIHT) and Embedded Block Coding with Optimal Truncation (EBCOT). The experiment results illustrate that the present invention can much more largely reserve texture information and small objects of the image in the same compression rate after being combined with the quantization model provided by the present invention. And the subjective effect of the present invention is better than original methods.