Combined Use of a Bilateral and Median Filter to Suppress Gaussian Noise in Images

Abstract

The paper is devoted to the problem of processing images with Gaussian noise. The method proposes a new way of denoising. This algorithm can effectively clean the image from noise and keep the edges of objects precise. The research was conducted in MATLAB R2019b, the quality of the proposed filter was tested using a peak signal-to-noise ratio and a structural similarity index. This method can de-noise images more efficiently in comparison with most existing methods. The proposed method de-noise better by 1-6 dB than comparable methods in the work. It can be used in applications, image processing, or as a precleaning program for an image recognition neural network.