Salt and pepper denoising filters for digital images: A technical review

• Published in: Serbian journal of electrical engineering Vol. 21; no. 3; pp. 429 - 466
• Main Authors: Kumar, Abhishek, Kumar, Sanjeev, Kar, Asutosh
• Format: Journal Article
• Language: English
• Published: Faculty of Technical Sciences in Cacak 01.01.2024
• Subjects: convolution neural networks; fuzzy logic; non-linear filter; peak signal-to-noise ratio; salt and pepper noise; structural similarity index measure
• ISSN: 1451-4869; 2217-7183; 2217-7183
• DOI: 10.2298/SJEE2403429K

Noise in images refers to random variations in pixel intensities that alter the original pixel intensities of the image. Among the various noises present in the image, salt and pepper noise corrupts images due to a defect in the device?s hardware or the camera?s faulty sensor. This leads to misinterpretation of pixels and deterioration of image quality during visualization of natural images and diagnosis of medical images. Up until now, researchers have presented several cutting-edge filters to overcome and lessen the impact of this noise. This article presents a comprehensive investigation into three different domains of impulse denoising of digital images. These domains are based on the spatial domain, the fuzzy logic domain, and the deep learning-based category. In this study, many techniques of image denoising were categorized and analyzed, along with their respective motivations, principles of execution, and comparative analysis. We carefully explain and implement a few significant approaches, considered state-of-the-art in each subject, in MATLAB. When doing simulations, the filters are analyzed and quantitatively evaluated using three metrics that are frequently utilized. These parameters are the peak signal-to-noise ratio (PSNR) and the structural similarity index measure (SSIM). Finally, we provide a comparison of each study category to enhance our comprehension of each domain. We conclude by outlining the challenges each domain poses and providing a detailed explanation of the rationale for future research.