Description:

This archive contains a Matlab implementation of the Multilinear Principal Component Analysis (MPCA) algorithm and MPCA+LDA, as described in the paper

Haiping Lu, K.N. Plataniotis, and A.N. Venetsanopoulos, "MPCA: Multilinear Principal Component Analysis of Tensor Objects", IEEE Transactions on Neural Networks, Vol. 19, No. 1, Page: 18-39, January 2008.

Algorithm 1: "MPCA.m" implements the MPCA algorithm described in this paper.

Algorithm 2: "MPCALDA.m" implements the MPCA+LDA algorithm in this paper.

%[Usages]% Please refer to the comments in the codes, which include example usage on 2D data and 3D data below:

FERETC80A45.mat: 320 faces (32x32) of 80 subjects (4 samples per class) from the FERET database

USF17Gal.mat: 731 gait samples (32x22x10) of 71 subjects from the gallery set of the USF gait challenge data sets version 1.7

%[Verification of gait recognition results]% To verify the gait recognition results presented in Table VII of the paper on a smaller version of the gait data in folder "USFGait17_32x22x10" so the numbers are not exactly the same

1. Run GRTestMPCA.m to get the results for ETG
2. Run GRTestMPCALDA.m to get the results for ETGLDA

testData.m specifies the data directory and probes to be processed

MADAll.m calculates the rank 1 and rank 5 identification rates using MAD measure (Table II) and symmetric matching.

GRResultsVerify.txt is the expected output in the command window.

%[Toolbox]% The code needs the tensor toolbox available at http://csmr.ca.sandia.gov/~tgkolda/TensorToolbox/

This package includes tensor toolbox version 2.1 for convenience.

%[Restriction]% In all documents and papers reporting research work that uses the matlab codes provided here, the respective author(s) must reference the following paper:

[1] Haiping Lu, K.N. Plataniotis, and A.N. Venetsanopoulos, MPCA: Multilinear Principal Component Analysis of Tensor Objects", IEEE Transactions on Neural Networks, Vol. 19, No. 1, Page: 18-39, January 2008.

Changes to previous version:

1. The MPCA paper is updated with a typo (the MAD measure in Table II) corrected.

2. Tensor toolbox version 2.1 is included for convenience.

3. Full code on gait recognition is included for verification and comparison.

BibTeX Entry: Download

Corresponding Paper BibTeX Entry: Download

Supported Operating Systems: Linux, Windows, Unix, Solaris

Data Formats: Matlab

Tags: Dimensionality Reduction, Pca, Feature Extraction, Principal Component Analysis, Multilinear Subspace Learning, Tensor

Archive: download here

Comments

No one has posted any comments yet. Perhaps you'd like to be the first?

Leave a comment

You must be logged in to post comments.