A pixel-domain post-processing technique to reduce the blocking artifacts in transform-coded images

摘要

The reconstructed images after compression produce annoying blocking artifacts near block boundaries, particularly in low bit-rate image compression, because each block is transformed and quantized independently Detection of blocking artifact has an important role in the design of image coding systems. Increasing the bit rate to obtain better quality images is too costly. In this paper, a novel pixel-domain technique for image blocking artifact detection is presented. The algorithm detects the regions of the image which present visible blocking artifacts. The blocking artifact detection and reduction is presented as a post- processing in the image decoder. The algorithm first detects the direction of artifact through power spectrum estimation of the e pixel absolute difference (PAD) in both horizontal and vertical directions. The power spectrum of the PAD is estimated using the Maximum Entropy method (MEM) which is a nonparametric method to perform accurate power spectrum estimation. The reduction process is applied on the direction of artifacts only to improve the image quality. The key of the reduction process is based on the modified projection operator to give reasonable modifications for the block boundary pixels. The simulation results illustrate the performance of the proposed method for the JPEG decoded images.