A Recursive Mean Filter for Image Denoising
Uğur Erkan, Serdar Enginoğlu, Dang N. H. Thanh · 2019 International Artificial Intelligence and Data Processing Symposium (IDAP) · 2019
In this article, we propose a Recursive Mean Filter (RMF) to remove the salt-and-pepper noise (SPN). We compare RMF with similar methods such as Decision Based Algorithm (DBA), Modified Decision-Based Unsymmetric Trimmed Median Filter (MDBUTMF), and Noise Adaptive Fuzzy Switching Median Filter (NAFSMF). In the experiments, we implement the test on 15 native images of the MATLAB library. To assess the denoising quality, we consider the following noise densities (noise levels): 10%, 30%, 50%, 70%, 90%; and use the well-known two quality metrics: Peak Signal-to-Noise Ratio (PSNR) and Structural Similarity (SSIM). The results show that RMF removes the noise effectively and competes with other state-of-the-art denoising methods.