A novel adaptive de-blocking algorithm for the smoothlets reconstructed image
Fang Xie, Chang Duan, Jianshu Cao · 2013
The classic multi-scale smoothlet transform is block oriented image approximations transform, which the source image will be divided into many distinct blocks. A major drawback related to this block division is that the reconstructed images exhibit highly noticeable discontinuities at the boundaries of blocks, namely blocking artifacts. In this paper, a novel adaptive de-blocking algorithm for the smoothlet reconstructed image is proposed which can reduce the boundaries' discontinuities according to the strength of blocking artifacts. From the experiments, the proposed method can effectively reduce the artifacts and improve the PSNR. And the improvement of PSNR is better than several conventional de-blocking methods based on weighted-sum.