BDCT compressed image deblocking usingweighted adaptive total variation
Wei Dai, Oscar C. Au, Wenjing Zhu, Xingyu Zhang, Feng Zou, Chao Pang · 2013
Images encoded at low bit rate usually exhibit visually annoying coding artifacts, which are commonly referred as blocking artifacts. In this paper, a novel weighted adaptive total variation method is proposed to remove the blocking artifacts. Based on the analysis of image coding process and block-based discrete cosine transform (BDCT) image properties, image deblocking is formulated as an optimization problem which is solved through approximating the objective function by a set of convex functions. Experimental results show that the proposed method can achieve better objective and subjective quality performance compared to other deblocking algorithms.