A novel deblocking algorithm using edge flow-directed filter and curvelet transform

A new post-processing approach based on "edge flow" and curvelet transform is proposed for the suppression blocking artifacts in block discrete cosine transform (BDCT) compressed images. Firstly, by exploiting the edge flow correlations, edge information in the compressed images is extracted and protected, while blocky noise in the smooth background regions is smoothed out by an edge flow-directed filter in the wavelet domain. Then, the curvelet transform coefficients in different subbands are filtered with adaptive thresholds that are obtained according to the edge flow boundary map. The advantage of the new method is that it retains sharp features in images and, compared with other wavelet-based methods, it is capable of achieving higher peak signal-to-noise ratio (PSNR) improvement as well as giving visually very pleasing images.