Comparison of Methods for Removing Noise in an Image Obtained in a Specular Microscope at a Wavelength of 13.84 nm
D. G. Reunov, И. В. Малышев, Н. И. Чхало · Journal of Surface Investigation X-ray Synchrotron and Neutron Techniques · 2022
Abstract— In this paper, selection of the optimal filtering method and variations in its parameters are carried out to improve the image quality without losing the useful signal. A new image-filtering method that does not lead to the blurring of images is proposed to remove noise. The appearance of noise is associated with heating of the recording system of the specular “soft” X-ray microscope. At the moment, there are a vast number of methods and algorithms for removing noise from an image. However, most the methods blur the original image because of filtering according to the filter kernel. This blurring leads to a deterioration in the resolution or even to the loss of some useful information, which can play an important role in the processing of experimental images. A combined image-filtering method that includes a threshold filter and a median one is proposed. This combination allows one to initially find local jumps or dips of the signal (the range of which is set manually), which will later be replaced by median values of the vicinity of a point in the image. As a result, the median filter is not applied to the entire image, but only where there are sharp differences in the intensity, so it does not blur the image as a whole.