Restoration of images corrupted by stripe interference using Radon domain filtering

The article deals with the problem of removing noise that has some anisotropy in a certain direction, in images received as a result of remote sensing. Such interference can occur with satellite imagery of the surface of Earth and planets due to the peculiarities of the imaging equipment. In the article, the method of removing such noise in the Radon space is considered, using its singular value decomposition. The use of this approach provides significant advantages over spatial filtering methods when pixel brightness values are used and can be a noticeable loss of useful information in the form of a blurring of borders. When using filtering in the Radon space to remove periodic noise predominantly only interference is removed, since only a small part of the Radon projections corresponds to noise. The numerical experiments on real-world images demonstrate the efficiency of the techniques proposed.