基于残差网络的图像缺失修复研究

Research on Image Missing Restoration Based on Residual Network

图像修复技术在古文物保护等领域具有重要的作用,并且是计算机视觉中有挑战性的任务之一。本文提出一种基于残差网络的图像修复算法,通过使用残差网络将多尺度特征信息进行融合学习,利用学习到的特征信息在图像修复过程,根据图像纹理走向和图像细节进行微调。结果显示,该算法可以有效地修复有缺损的图像,修复后的图像可以最大程度地减少图像失真的情况。

Image restoration technology plays an important role in the field of ancient cultural relics protection,and it is one of the challenging tasks in computational vision.This paper proposes an image restoration algorithm based on residual network.The multi-scale feature information is fused and learned by using residual network.The learned feature information is used to fine tune the image according to the image texture direction and image details in the image restoration process.The results show that the algo⁃rithm can effectively repair the defective image.The repaired image can minimize the image distortion.