文献类型：期刊文献

中文题名：基于结构自动匹配的仿射相似破损图像修复

英文题名：The inpainting of affine similarity damaged image based on structure automatic match

作者：于文静[1];毕东旭[1];颜学峰[1]

机构：[1]华东理工大学信息科学与工程学院,上海200237

年份：2017

卷号：52

期号：3

起止页码：32

中文期刊名：山东大学学报（理学版）

外文期刊名：Journal of Shandong University(Natural Science)

收录：CSTPCD;;北大核心:【北大核心2014】；CSCD:【CSCD2017_2018】；

语种：中文

中文关键词：仿射相似;结构稀疏算子;自动匹配

外文关键词：affine similarity; structure sparsity operator; automatic match

摘要：基于图像结构稀疏性定义了图像的结构稀疏算子,利用稀疏算子实现原图像到结构图像的映射。根据仿射相似图像具有相同结构的偏移量成稀疏分布的特点,统计相同结构偏移量的分布特征,获得破损区域的最优匹配信息。实验结果表明,该算法可以实现结构的自动匹配,在仿射破损图像的修复中更有效。
Based on the character of image structure sparsity, the image structure sparsity operator was defined which can map from original image to structure image. According to the property of the sparse distribution of the similar structure offsets in affine similarity image, this paper indicates the distribution character of similar structure offsets through scien- tific statistics, with the objective of adaptively winning the superior matching region to the broken region. Experimental results show that the algorithm can achieve the structure match adaptively and can complete the affine broken image ef- fectively.