摘要
提出了一种全局一致性和局部连续性结合的壁画修复算法.主要利用线性系统和图像修复间所蕴含的关系,构建具有全局过完备特性的相似块集合字典,同时构造弹性网正则化下的稀疏修复模型,并用同伦-最小角回归法求解出过完备字典下的稀疏系数;通过字典与系数的线性组合,得到待修复区域的全局特征;之后提出一种基于领域相似特性的局部特征估计方法,得到待修复区域的局部特征;最后对上述得到的全局和局部特征信息做线性加权,完成整个待修复区域的填充.实验结果表明,本算法能很好地解决相关修复算法在修复时所出现的纹理错误填充、结构不连续以及“块效应”等现象,并且得到较好的修复结果 .
This paper proposes a global uniform and local continuity repair method for mural image inpainting.It uses the relationship between linear system and image repair to construct the similarity-preserving overcomplete dictionary with global weighted feature. Meanwhile, a novel sparse repair model with elastic net regularization based on similarity-preserving overcomplete dictionary is formulated to enhance the global feature consistency, and then an estimated method of neighborhood similarity is presented to guarantee local feature consistency, finally, a global feature patch and local feature patch weighted method are applied to obtain the target patch. Experimental results on damaged murals demonstrate the proposed method outperforms state-of-the-art inpainting methods.
作者
王欢
李利
李庆
邓筠钰
商惠敏
WANG Huan;LI Li;LI Qing;Deng Junyu;SHANG Huiming(Industrial Technology Research Center,Guangdong Institute of Scientific&Technical Information,Guangzhou 510033,China;College of Civil and Transportation Engineering,Shenzhen University,Shenzhen 518061,China;School of Automation Science and Engineering,South China University of Technology,Guangzhou 510641,China)
出处
《湖南大学学报（自然科学版）》
EI
CAS
CSCD
北大核心
2022年第6期135-145,共11页
Journal of Hunan University:Natural Sciences
基金
广东省重点领域研发计划项目(2018B010112001)
广东省科技计划项目(XMHT-202103008)。
关键词
线性系统
敦煌壁画
稀疏表示
图像修复
linear systems
dunhuang murals
sparse representation
image inpainting
