Outline:
Folder structure
Usage
Citation
URL of third-party toolboxes and functions
Folder structure:
data\ : example data
|-- testMSI_1.mat : example MSI data 1
|-- testMSI_2.mat : example MSI data 2
|-- testVideo.mat : example vedio data
|-- groundtruth_at_16th_frame.bmp : groundtruth of vedio data at 16th frame
lib\ : library of code
|-- tensor_toolbox\ : toolbox for tensor operations [1]
|-- my_tensor_toolbox\ : another toolbox for tensor operations
|-- KBR\ : subfunctions of KBR method
|-- quality_assess\ : functions of quality assessment indices
| |-- ssim_index.m : SSIM [2]
| |-- FeatureSIM.m : FSIM [3]
| |-- ErrRelGlobAdimSyn.m : ERGAS
| |-- SpectAngMapper.m : SAM
| |-- MSIQA.m : interface for calculating PSNR and the four indices above
|-- compete_methods : competing methods (down loaded or implemented based on reference papers)
| |-- HOrpca\ : HOrpca toolbox [4]
| |-- inexact_alm_rpca\ : matrix ADMM-based rpca toolbox [5]
| |-- tensor_SVD\ : tSVD toolbox [6]
| |-- TMac\ : TMac toolbox [7]
| |-- TenCompletion\ : McpTC and ScadTC toolbox [8]
| |-- LRTV_TC.m : functions for Trace/TV method [9]
| |-- Mc_adm.m : functions for MC_ALM method [10]
| |-- Tc_liu.m : functions for HaLRTC method [11]
|-- togetGif.m : function for getting GIF result
|-- showVideoResult.m : scripts that show the result of background subtraction experiment
|-- showMSIResult.m : scripts that show the result of MSI completion experiment
Demo_TC_MSI.m : scripts that applies the tensor-completion methods and calculates the QA indices
Demo_RPCA_video.m : scripts that applies the robust-principal-component-analysis methods and calculates the QA indices
KBR_TC.m : function for sloving the intrinsic tensor sparsity based tensor completion model
KBR_RPCA.m : core function of the the intrinsic tensor sparsity based tensor RPCA model
Usage:
(1) For MSI completion experiment, you can simply follow these steps: 1.Re-arrange the MSI into [0, 1]. 2.Add the folder 'lib'into path, and use the function KBR_TC as follows: [ restored_img ] = KBR_TC( courrupted_img, Omega) Please type 'help KBR_TC ' to get more information.
You may find example codes in file Demo_TC_MSI.m
Also, you can use the demo to see some comparison. You can:
1. Type 'Demo_TC_MSI' to to run various methods and see the pre-computed results.
2. Use 'help Demo_TC_MSI' for more information.
3. Change test MSI by simply modifying variable 'dataname' in Demo_TC_MSI.m (NOTE: make sure your MSI
meets the format requirements).
4. Change sampling rate by modifying variables 'sample_ratio ' in Demo_TC_MSI.m
5. Select competing methods by turn on/off the enable-bits in Demo_TC_MSI.m
(2) For background subtraction experiment, you can simply follow these steps: 1.Re-arrange the video into [0, 1]. 2.Use the function KBR_RPCA as follows: [ restored_img ] = KBR_RPCA( tensorData,beta,gamma) Please type 'help KBR_RPCA ' to get more information.
You may find example codes in file Demo_RPCA_video.m
Also, you can use the demo to see some comparison. You can:
1. Type 'Demo_RPCA_video' to to run various methods and see the pre-computed results.
2. Use 'help Demo_RPCA_video' for more information.
3. Change test video data by simply modifying variable 'dataname' in Demo_RPCA_video.m (NOTE: make sure your video
meets the format requirements).
4. Select competing methods by turn on/off the enable-bits in Demo_RPCA_video.m
Citation:
Qi Xie, Qian Zhao, Deyu Meng*, & Zongben Xu
Kronecker-Basis-Representation Based Tensor Sparsity and Its Applications to Tensor Recovery[J].
IEEE Transactions on Pattern Analysis & Machine Intelligence, 2017, PP(99):1-1. (accepted)
BibTeX:
@article{Qi2017Kronecker,
title={Kronecker-Basis-Representation Based Tensor Sparsity and Its Applications to Tensor Recovery},
author={Qi, Xie and Qian, Zhao and Meng, Deyu and Xu, Zongben},
journal={IEEE Transactions on Pattern Analysis & Machine Intelligence},
volume={PP},
number={99},
pages={1-1},
year={2017},
}
URL of the toolboxes and functions and citation of competing methods:
[1] tensor_toolbox http://www.sandia.gov/~tgkolda/TensorToolbox/index-2.5.html
[2] ssim_index.m https://ece.uwaterloo.ca/~z70wang/research/ssim/
[3] FeatureSIM.m http://www4.comp.polyu.edu.hk/~cslzhang/IQA/FSIM/FSIM.htm
[4] HOrpca https://sites.google.com/site/tonyqin/research
[5] inexact_alm_rpca http://perception.csl.illinois.edu/matrix-rank/home.html
[6] t-SVD http://www.ece.tufts.edu/~shuchin/software.html
[7] Tmac http://www.caam.rice.edu/~yx9/TMac/
[8] W. Cao, Y. Wang, C. Yang, X. Chang, Z. Han, and Z. Xu. Foldedconcave penalization approaches to tensor completion.
Neurocomputing, 152:261¨C273, 2015.
[9] M. Golbabaee and P. Vandergheynst. Joint trace/tv norm minimization: A new efficient approach for spectral compressive
imaging. pages 933¨C936, 2012.
[10] Z. Lin, M. Chen, and Y. Ma. The augmented lagrange multiplier method for exact recovery of corrupted low-rank matrices.
Eprint Arxiv, 9, 2010.
[11] J. Liu, P. Musialski, P. Wonka, and J. Ye. Tensor completion for estimating missing values in visual data. IEEE Trans.
Pattern Analysis & Machine Intelligence, 35(1):208¨C220, 2013.