This repository provides the Matlab codes of the paper entitled 'Diffusion Based Method for Impulse Noise Removal Using Residual Feedback' (https://www.sciencedirect.com/science/article/abs/pii/S0898122121004569)
In the diffusion based processes, the gradient operator performs well in the case of Gaussian noise and can distinguish between edges (large gradient) and the noise-contaminated areas, while impulse noise can cause large gradients and consequently be preserved like the edges. Therefore, in this paper, the ENI operator which distinguishes well between impulse noise and signal is analyzed and using it, an efficient diffusion based method for impulse denoising is proposed. The proposed method is carried out in 2 phases using the residual feedback and the localization of the noisy pixels.The experimental results confirm the efficiency of the proposed method according to subjective and objective criteria.
Matlab
To run the codes, please simply run Main.m
[1] Rafsanjani, Hossein Khodabakhshi, Hossein Noori, and Nasibe Naseri. "Diffusion based method for impulse noise removal using residual feedback." Computers & Mathematics with Applications 107 (2022): 45-56.