Eficient image denoising using deep learning: A brief survey

Bo Jiang, Jinxing Li, Yao Lu, Qing Cai, Huaibo Song, Guangming Lu

Abstract: Image denoising is a vital computer vision task that aims to remove noise from images. Deep learning techniques have made remarkable progress in this field in recent years. This survey provides a comprehensive overview of efficient deep learning-based image denoising methods. Unlike previous reviews, it focuses exclusively on models based on efficient deep learning and examines the latest developments in image denoising from a unique deep learning perspective. This includes datasets, metrics, model design, and learning strategies. First, we review the definition of image denoising problems and introduce mainstream natural image denoising datasets and metrics. Then, we conduct a comprehensive review of the two mainstream frameworks (straight-tube and U-shaped) for image denoising, analyzing their advantages, limitations, and distilling basic principles or design strategies. We also summarize and analyze different learning strategies for efficient image denoising. Finally, the survey summaries current challenges and future directions, providing insightful guidance for future research in deep learning-based efficient image denoising.