The Rebuilding of Texture Information of Compressed Image Based on JPEG2000
Liyu Lin · Journal of Wuhan University · 2006
Some details of the image will lose in the image lossy compression.Occasionally,these details,such as the texture of the human skin, will make an image look unnaturally though the real distortion of the image is not sever very much.This paper proposes a new method of rebuilding the statistical texture in compressed image which is based on JPEG2000 in order to get better subjective compensation by less cost of bit-rate.Meanwhile compared with non-linear extended quantization methods' in both quality parameter and subjective effect,the method make it possible to regenerate the texture for the decoded image.By encoding the statistical result of the texture information of the compressed image during the coding process,which is on the assumption that the Gauss distributing is effective here.