Advances in denoising by multiple copies

Conventional image restoration algorithms use transform-domain filters, which separate noise from the sparse signal among the transform components or apply spatial smoothing filters in real space whose design relies on prior assumptions about the noise statistics. These filters also reduce the information content of the image by suppressing spatial frequencies or by recognizing only a limited set of shapes. We have previously proposed a new method called MC-MLP1 (multiple copies, multiple layer perceptrons), where we have shown that denoising can be efficiently done with a nonlinear filter that operates along patch neighborhoods and multiple copies of the original image. The use of patches enables the algorithm to account for spatial correlations in the random field whereas the multiple copies are used to recognize the noise statistics. The nonlinear filter, which is implemented by a hierarchical multistage system of multilayer perceptrons (MLP), outperformed state-of-the-art denoising algorithms such as those based on collaborative filtering and total variation. Compared to conventional denoising algorithms, our method can restore images without blurring them, making it attractive for use in applications such as medical imaging where, the preservation of anatomical details is critical.