At low bit-rate, most block-based image compression schemes generate artifacts known as block effect. Many existing approaches concentrated on concealing the discontinuity around the block boundary regions to produce a visually more pleasant image with no or little improvement in PSNR. We proposed a novel approach which attempts to recover the DC components of each block from the coded ones in order to calibrate the reference level for all the pixels within the block. The calibration of DC components starts from two anchor blocks centered in the image with their DC components coded error free. It then proceeds in a spiral scan fashion to exploit the maximum constraints imposed by the already updated neighboring blocks. After DC components calibration, a Huber-Markov random field model based POCS algorithm is applied to obtain good reconstruction both in terms of PSNR improvement and visual observation.
|Original language||English (US)|
|Journal||ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings|
|State||Published - Jan 1 1994|
ASJC Scopus subject areas
- Signal Processing
- Electrical and Electronic Engineering