Reduction of blocking artifacts for HDTV using offset-and-shift technique

Abstract

This paper addresses an effective post-processing technique to remove blocking artifacts, which severely degrade visual quality in block-based image coding. The proposed algorithm classifies blocks into four types based on directional activity and removes the blocking artifacts while preserving edges and texture information by applying different types and strength of filter for each block type. Moreover, the proposed scheme includes a two-dimensional filtering method in order to remove staircase noises and corner outliers around diagonal and curved edges. Experimental results show that the proposed algorithm outperforms a number of deblocking methods in the literature in terms of PSNR and subjective visual quality. Furthermore, the execution time of the proposed algorithm takes less than that of the other methods, and thus it can be incorporated into real-time applications for moving pictures as well as still images.