Adaptive blocking artifact reduction using wavelet-based block analysis

In this paper a signal adaptive postprocessing algorithm is presented to remove visually annoying blocking artifacts in BDCT-coded images. The wavelet-based block analysis is adopted to exploit the directional activity of each block, as a systematic approach in decision of the block type. The proposed 1D filters that are applied activity-adaptively alleviate grid noises effectively and a directional 2D filter reduces staircase noises and corner outliers near diagonal edges while preserving original details. Comprehensive experiments indicate that the proposed algorithm outperforms a number of deblocking methods in the literature in terms of PSNR and subjective visual quality. Moreover, the proposed algorithm takes less run time than those of the conventional methods, and thus it can be used in real-time applications for moving pictures as well as still images.