摘要
为了让信息从周围向待修复区域填充时保持图像边缘结构,提出结构保持图拉普拉斯正则的图像修复模型.该模型在信号依赖的图拉普拉斯矩阵基础上,引入梯度图像平滑约束,能够促进目标函数的最优解收敛至分片平面图像;此外,该模型转化为无约束二次规划问题,可以通过共轭梯度法快速求解.实验结果表明,所提的图像修复算法相比于现有图像修复算法不仅速度快,而且可以克服图拉普拉斯正则图像修复算法所产生的块效应,使得复原后的图像更加自然.
A structure-preserving graph Laplacian regularizer is proposed for image inpainting to improve structure-preserving of image edges while information propagation from the surrounding known regions to the unknown.Based on the signal-dependent graph Laplace matrix,the model introduces gradient image smoothness to promote the objective function to converge to the piecewise planar images.The inpainting model is transformed into an unconstrained quadratic programming problem,which can be solved quickly by the conjugate gradient method.The experimental results show that the speed of the proposed algorithm is not only faster than the existing inpainting algorithms,but also can restore images which look less blocky and more natural than traditional Laplacian regularizer.
作者
曾勋勋
陈飞
ZENG Xunxun;CHEN Fei(School of Mathematics and Statistics,Fuzhou University,Fuzhou,Fujian 350108,China;College of Computer and Data Science,Fuzhou University,Fuzhou,Fujian 350108,China)
出处
《福州大学学报（自然科学版）》
CAS
北大核心
2022年第3期323-329,共7页
Journal of Fuzhou University(Natural Science Edition)
基金
国家自然科学基金资助项目(61771141)
福建省教育厅中青年教师教育科研项目(JAT190020)。
关键词
图像修复
图信号处理
拉普拉斯矩阵
分片平滑
结构保持
image inpainting
graph signal processing
Laplacian matrix
piecewise smoothness
structure-preserving
