FAST & ROBUST IMAGE INTERPOLATION USING GRADIENT GRAPH LAPLACIAN REGULARIZER

被引:13
作者
Chen, Fei [1 ]
Cheung, Gene [2 ]
Zhang, Xue [2 ]
机构
[1] Fuzhou Univ, Coll Math & Comp Sci, Fuzhou, Peoples R China
[2] York Univ, Dept EECS, Toronto, ON, Canada
来源
2021 IEEE INTERNATIONAL CONFERENCE ON IMAGE PROCESSING (ICIP) | 2021年
基金
中国国家自然科学基金;
关键词
Image interpolation; graph signal processing; SPARSE;
D O I
10.1109/ICIP42928.2021.9506064
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
In the graph signal processing (GSP) literature, it has been shown that signal-dependent graph Laplacian regularizer (GLR) can efficiently promote piecewise constant (PWC) signal reconstruction for various image restoration tasks. However, for planar image patches, like total variation (TV), GLR may suffer from the well-known "staircase" effect. To remedy this problem, we generalize GLR to gradient graph Laplacian regularizer (GGLR) that provably promotes piecewise planar (PWP) signal reconstruction for the image interpolation problem-a 2D grid with random missing pixels that requires completion. Specifically, we first construct two higher-order gradient graphs to connect local horizontal and vertical gradients. Each local gradient is estimated using structure tensor, which is robust using known pixels in a small neighborhood, mitigating the problem of larger noise variance when computing gradient of gradients. Moreover, unlike total generalized variation (TGV), GGLR retains the quadratic form of GLR, leading to an unconstrained quadratic programming (QP) problem per iteration that can be solved quickly using conjugate gradient (CG). We derive the means-square-error minimizing weight parameter for GGLR, trading off bias and variance of the signal estimate. Experiments show that GGLR outperformed competing schemes in interpolation quality for severely damaged images at a reduced complexity.
引用
收藏
页码:1964 / 1968
页数:5
相关论文
共 31 条
[1]   ON THE RATE OF CONVERGENCE OF THE PRECONDITIONED CONJUGATE-GRADIENT METHOD [J].
AXELSSON, O ;
LINDSKOG, G .
NUMERISCHE MATHEMATIK, 1986, 48 (05) :499-523
[2]   Graph-Based Blind Image Deblurring From a Single Photograph [J].
Bai, Yuanchao ;
Cheung, Gene ;
Liu, Xianming ;
Gao, Wen .
IEEE TRANSACTIONS ON IMAGE PROCESSING, 2019, 28 (03) :1404-1418
[3]  
Bigdeli SA, 2017, ADV NEUR IN, V30
[4]  
Borenstein E., 2002, EUR C COMP VIS ECCV
[5]   A TGV-Based Framework for Variational Image Decompression, Zooming, and Reconstruction. Part I: Analytics [J].
Bredies, Kristian ;
Holler, Martin .
SIAM JOURNAL ON IMAGING SCIENCES, 2015, 8 (04) :2814-2850
[6]   Total Generalized Variation [J].
Bredies, Kristian ;
Kunisch, Karl ;
Pock, Thomas .
SIAM JOURNAL ON IMAGING SCIENCES, 2010, 3 (03) :492-526
[7]   Image recovery via total variation minimization and related problems [J].
Chambolle, A ;
Lions, PL .
NUMERISCHE MATHEMATIK, 1997, 76 (02) :167-188
[8]   Bias-Variance Tradeoff of Graph Laplacian Regularizer [J].
Chen, Pin-Yu ;
Liu, Sijia .
IEEE SIGNAL PROCESSING LETTERS, 2017, 24 (08) :1118-1122
[9]   Robust Semisupervised Graph Classifier Learning With Negative Edge Weights [J].
Cheung, Gene ;
Su, Weng-Tai ;
Mao, Yu ;
Lin, Chia-Wen .
IEEE TRANSACTIONS ON SIGNAL AND INFORMATION PROCESSING OVER NETWORKS, 2018, 4 (04) :712-726
[10]   Graph Spectral Image Processing [J].
Cheung, Gene ;
Magli, Enrico ;
Tanaka, Yuichi ;
Ng, Michael K. .
PROCEEDINGS OF THE IEEE, 2018, 106 (05) :907-930