Adaptive reduction of blocking artifacts in DCT domain for highly compressed images

Abstract

A major drawback of block discrete cosine transform-based compressed image is the appearance of visible discontinuities along block boundaries in low bit-rate coding, which are commonly referred as blocking artifacts. Reduction of blocking artifacts can significantly improve image quality and extend maximum compression ratio under certain subjective quality. In this paper, a method for blind measuring the degree of blocking artifacts and postprocessing technology for removing these discontinuities are proposed, and it is most important that all these works are operated in DCT domain. By Walsh transform and local threshold technology, the precision of edge detection by Sobel operator is improved, so blurring in objective edge is reduced. In the process of deblocking, compensatory coefficients matrices and adaptive smooth method are introduced to improve postprocessed image performance. Experiments demonstrate that the combination of all these methods gives excellent results compared with other approaches before in both subjective and objective viewpoints.