Dear Lamis Ghoualmi,

Look the link. May be useful.

http://class.inrialpes.fr/pub/202-tan-amfg07b.pdf

http://stackoverflow.com/questions/26681081/feature-reduction-within-gabor-filter-banks-deploying-pca

Regards, Shafagat

Dear kouser , my data is more biger than that it is just an exemple

Thank you for yours answer guys

I work with matlab,  for exemple:  [eigVector eigValue]=pca(DATA) 

how i calculate the new DATA using eigVector and EigValue (projection of  the data)

Hi Lamis

The projected data:(new data) Y=X.V where vi are the eigenvectors corresponding to the largest m eigenvalues.

Dear M ouadfel,

Thank you for your response,  please  i need more informations

[eigvector eigvalue]=pca(data);

NewData=eigvector * Data   but the dimension  mismatch ?

Hi Lamis

You can also use transpose of 'eigvector '

I  suggest you this link

http://www.visiondummy.com/2014/05/feature-extraction-using-pca/

You can try to contact a mister Mohammed Waleed Ashour. His email is at ashour1234 AT yahoo DOT com. He used PCA to reduced gabor features, the same as what you are looking to do. Perhaps he can provide the code, as he did it in MATLAB as well. 

Minitab has support for PCA, and is relatively simple to use. In my opinion it is the easiest way to apply PCA.

You can see simple instructions here:

http://support.minitab.com/en-us/minitab/17/topic-library/modeling-statistics/multivariate/principal-components-and-factor-analysis/perform-pca-with-varimax-rotation/

Hi ,

please check the below link, it may usefull

http://in.mathworks.com/help/stats/princomp.html

http://www.nlpca.org/pca-principal-component-analysis-matlab.html

http://matlabdatamining.blogspot.in/2010/02/principal-components-analysis.html

https://www.cs.princeton.edu/picasso/mats/PCA-Tutorial-Intuition_jp.pdf

http://luthuli.cs.uiuc.edu/~daf/courses/CS-498-DAF-PS/Lecture%209%20-%20PCA.pdf

Hi,

you should change your matrix dimension and sit every picture one rows then apply the PCA to it.

so you have the matrix which has two dimention. num of row = num of pictures in database and num of column = features of the picture that you want to reduce

then apply the PCA to it so you can get a matrix with less column than it before.

it is necessary to divided the matrix into training (U_training ) and test (U_test) then apled PCA in training (U=PCA(U_training,N)),

where: N is the numbre of features extractionU the matrix of projection

reduce Gabor features=U'*U_test