Designer: Junbo Zhao, Wuhan University, Working in Tsinghua National lab of intelligent images and documents processing.
Email: [email protected] +86-18672365683

Introduction:
This package includes a small and convenient lib for Matrix operations, and a Principle Component Analysis (PCA) program. This PCA project is designed to be suitable to High Dimensional situations, using some linear algebra tricks to transform the covariance matrix and achieving a faster and less memory-cost way for PCA.

Platform:
This program is tested on VS2010, 32bit, Win7 system. I cannot guarantee it could be adopted on other platforms.
For Windows users, you can just add the .cpp and .h files in your project and compile it.
For Linux users, I will soon update this program with a makefile.
For Mac users, I am sorry that this program is and will not be tested on Mac system. If you can successfully compile this program on Mac, please feel free to contact me！

Files:
In detail, "eigen.cpp" as well as "eigen.h" combined finishes computing eigenvalues and corresponding eigenvectors;
"matrix.cpp" and "matrix.h" in charge of all the matrix operations;
"pca.cpp" implements PCA.

Usage:

1. You should prepare your .txt file of the feature matrix before implementation.
   The specific rules of .txt file can be found in the sample.txt.
2. View the main function (in pca.cpp) to see how to exploit PCA.