计算机应用与软件Issue(10):216-219,4.DOI:10.3969/j.issn.1000-386x.2014.10.052
一种改进的基于样例的图像修复算法
AN IMPROVED EXEMPLAR-BASED IMAGE INPAINTING METHOD
张绚 1尹立新 1李振杰1
作者信息
- 1. 山东大学信息科学与工程学院 山东 济南250100
- 折叠
摘要
Abstract
Classic Criminisi image inpainting algorithm inpaints both the texture and structure of the image simultaneously by combining the advantages of partial differential equation and texture synthesis algorithm,so gets better visual effect.But this exemplar-based image inpainting algorithm takes the confidence and the data term into consideration only while defining the priority,which leads to the propagation of error information,and attenuates the inpainting results.Based on Criminisi algorithm,we propose an improved exemplar-based image inpainting scheme.By redefining the priority of the inpainting patches,the scheme adds a geometrical parameter that measures the smoothness of image and enlarges the limitation condition of optimal matching patch at the same time,so that the inpainted images attains better visual effects.关键词
图像修复/样本/优先权/平滑性Key words
Image inpainting/Exemplar/Priority/Smoothness分类
信息技术与安全科学引用本文复制引用
张绚,尹立新,李振杰..一种改进的基于样例的图像修复算法[J].计算机应用与软件,2014,(10):216-219,4.