From Noise to Clarity: A Hybrid Approach for Image Denoising Using Traditional and Deep Learning Methods

Abstract

In this study, we explore various image denoising techniques to restore images affected by noise, with a particular focus on traditional and deep learning-based methods. The research compares conventional denoising approaches, including Wavelet Thresholding, Bilateral Filtering, Non-Local Means, and Wavelet Denoising with Bayesian Shrinkage, against state-of-the-art deep learning models, such as DnCNN and U-Net. The performance of these methods is evaluated based on two metrics: Peak Signal-to-Noise Ratio (PSNR) and Structural Similarity Index (SSIM). Additionally, we investigate the potential of Noise2Noise, a deep learning technique trained without clean images, to enhance the robustness of denoising in adverse weather conditions. The results indicate the strengths and weaknesses of both conventional and deep learning-based approaches, providing insights into their applicability in real-world image restoration tasks.