Image deblocking using convex optimization with modified total variation method

Images encoded at low bit rate usually exhibit visually annoying coding artifacts, commonly referred as the blocking artifacts. The reason for causing blocking artifacts is that each 8×8 block is quantized and encoded independently, which results in block discontinuities between neighboring blocks. In this paper, a convex optimization problem is formulated to deal with the blocking artifacts. Total variation (TV) method is introduced in this paper to impose the continuity constraint on the block boundaries. Moreover, in order to remove the staircase effect caused by the TV method, a modified TV method is proposed. Experimental results show that the proposed algorithm can remove the blocking artifacts effectively compared to other image deblocking algorithms.