图像修复研究进展综述

摘 要: 图像修复是指将信息缺失的图像进行像素填充以达到人类视觉满意的效果。文章对该领域研究的相关技术进行综述:首先描述了传统的图像修复方法,接着描述了基于深度学习的图像修复方法,介绍各类方法的适用范围及优缺点,最后对未来的研究方向及重点提出展望。
关键词: 图像修复; 深度学习; 卷积神经网络; 自编码网络; 生成式对抗网络
中图分类号:TP391.4 文献标识码:A 文章编号:1006-8228(2021)12-06-04
Abstract: Image restoration is to fill the missing image with pixels to achieve the satisfactory effect of human vision. This paper reviews the related technologies in this field: firstly, it describes the traditional image restoration methods, and then describes the image restoration methods based on deep learning, introduces the application scope, advantages and disadvantages of various methods, and finally puts forward the prospect of future research direction and focus.
Key words: image restoration; deep learning; convolutional neural network; self-coding network; generative adversarial network
0 引言
生活离不开图像,图像作为反映客观世界的重要载体,是人类接收外界信息的重要来源和手段。