Deep Learning-based Face Super-resolution: A Survey

Face super-resolution (FSR), also known as face hallucination, which is aimed at enhancing the resolution of low-resolution (LR) images to generate high-resolution images, a domain-specific image problem. Recently, FSR has received considerable attention and witnessed dazzling advances with development deep learning techniques. To date, few summaries studies on learning-based are available. In this survey, we present comprehensive review methods in systematic manner. First, summarize problem formulation introduce popular assessment metrics loss functions. Second, elaborate facial characteristics datasets used FSR. Third, roughly categorize existing according utilization characteristics. each category, start general description design principles, an overview representative approaches, then discuss pros cons among them. Fourth, evaluate performance some state-of-the-art methods. Fifth, joint other tasks, FSR-related applications introduced. Finally, envision prospects further technological advancement field.