Image denoising: who is the best?

• Main Author: Toh, Sheng Rong
• Other Authors: Qian Kemao
• Format: Final Year Project
• Language: English
• Published: Nanyang Technological University 2022
• Subjects: Engineering::Computer science and engineering
• Online Access: https://hdl.handle.net/10356/156546
• Institution: Nanyang Technological University

While high-quality images are often desirable, image noise is often inevitable. With that said, many image denoising methods have been developed over the years, and we want to compare and find the best image denoising method available for real-world images. We will be implementing traditional methods such as the non-local means (NLM) and block-matching and 3D filtering (BM3D), and deep learning models such as autoencoder, denoising convolutional neural network (DnCNN) and real image denoising with feature attention (RIDNet) for comparison. 160 coloured clean-noisy image pairs will be used in this experiment. Through this experiment, we have found that RIDNet is the most effective image denoising method out of the 5 mentioned above.