基于多管秩和全变分的低秩张量填充  

Low rank tensor completion based on multi-tubal rank and TV regularization

作　　者：马娇 汪丽琴 喻高航 MA Jiao;WANG Liqin;YU Gaohang(School of Sciences,Hangzhou Dianzi University,Hangzhou Zhejiang 310018)

机构地区：[1]杭州电子科技大学理学院,浙江杭州310018

出　　处：《杭州电子科技大学学报（自然科学版）》2023年第5期96-102,共7页Journal of Hangzhou Dianzi University：Natural Sciences

基　　金：国家自然科学基金资助项目(12071104)。

摘　　要：提出了一种基于酉变换张量低多管秩分解的张量填充方法。受多光谱图像中空间光谱平滑性特征的启发,在模型中加入了空间平滑(TV)正则项,并使用邻近交替极小化(PAM)算法求解该模型。多光谱图像(MSI)数据上的仿真实验表明所提模型和算法是有效的。In this paper,a new tensor completion model is proposed,which is based on multi-tubal rank and unitary transform.Inspired by the spatial characteristic of multispectral data,we incorporate a spatial-spectral smoothness regularization into the proposed model.Furthermore,we develop an efficient proximal alternating minimization algorithm to solve the problem.Extensive experiments on multispectral imagery(MSI)data shows that the proposed model and algorithm are effective.

关 键 词：张量填充 张量积 酉变换 数据恢复 

分 类 号：O221.2[理学—运筹学与控制论]