A Comprehensive Review of Deep Learning-Based Real-World Image Restoration

Real-world imagery does not always exhibit good visibility and clean content, but often suffers from various kinds of degradations (e.g., noise, blur, rain drops, fog, color distortion, etc.), which severely affect vision-driven tasks image classification, target recognition, tracking, etc.). Thus, restoring the true scene such degraded images is significance. In recent years, a large body deep learning-based processing works has been exploited due to advances in neural networks. This paper aims make comprehensive review real-world restoration algorithms beyond. More specifically, this provides overviews critical benchmark datasets, quality assessment methods, four major categories i.e., based on convolutional network (CNN), generative adversarial (GAN), Transformer, multi-layer perceptron (MLP). The highlights latest developments each category architecture provide an up-to-date overview. Moreover, representative state-of-the-art methods are compared visually numerically. Finally, for restoration, current situations objectively assessed, challenges discussed, future directions trends presented.