Image denoising using learned dictionary based on double sparsity model
Ruihua Liang, Zaixin Zhao, Shengguo Li · 2011
A novel image denoising algorithm is proposed. We introduce a new effective scheme to train a redundant dictionary from the noisy image itself. The scheme combines the double sparsity model and the zero-tree structure in the wavelet domain. The training vectors are constructed by regrouping the wavelet coefficients of high-frequency subbands in the same orientation across different scales. This scheme overcomes the limit on the input signal dimension as well as the over-fitting problem. We demonstrate the potential of this denoising algorithm with several experiments. The performance of our approach is competive to some state of the art denoising methods in some cases.