Applying Human Visual Model to JPEG2000 Codec

碩士 === 國立雲林科技大學 === 電機工程系碩士班 === 97 === Recently, the wavelet based JPEG2000 algorithm has attracted significant attentions from researchers in image coding. JPEG2000 maximizes the PSNR value (or minimize MSE value) for each transmitted bit. However, it is well known that PSNR is not a good indicato...

Full description

Bibliographic Details
Main Authors: Ruei-Cheng Jian, 簡瑞成
Other Authors: Day-Fann Shen
Format: Others
Language:zh-TW
Published: 2009
Online Access:http://ndltd.ncl.edu.tw/handle/41042284591357188056
Description
Summary:碩士 === 國立雲林科技大學 === 電機工程系碩士班 === 97 === Recently, the wavelet based JPEG2000 algorithm has attracted significant attentions from researchers in image coding. JPEG2000 maximizes the PSNR value (or minimize MSE value) for each transmitted bit. However, it is well known that PSNR is not a good indicator for the perceptual image fidelity. Human visual perception is more sensitive to distortions in lower frequency components than higher frequency components. In this paper, we enhance JPEG2000 by introducing a process between wavelet transform and JPEG2000 called JND_SQ. JND_SQ allowed JPEG2000 to maximize the perceptual fidelity instead of PSNR for each transmitted bit. We proposed a perceptual fidelity criterion called JND_PSNR based on the SY JND model. The new criterion is more consistent with human perception than PSNR. The experiment results show that JND_JPEG2000 are better than JPEG2000 in terms of JND_PSNR. The JND_SQ process can efficiently quantize the wavelet coefficient to zeros, which can be further utilized to improve the speed of JPEG2000. Finally, JND_SQ can be turned off if the original JPEG2000 is desired.