Feature selection code with input features matrix for 1725 images

I did feature extraction for 1725 images by Gabor transform with 40 kernels. The result features matrix is M with size 1725 * 100000 where size of each image is 50x50 pixels. So 40 times 50*50 = 100,000.

Now I have matrix M with 100000 columns * 1725 rows where each row of M matrix represent one of the 1725 images.

I used matlab program

Now i need making feature selection code using matlab where the input matrix is the previous M matrix.

Thanks

alyaa aly

Answer 1

[Image Analyst]

I understand that you took 1725 50x50 images and ran 40 Gabor filters on them and took all the pixels and strung them together onto one row of M.

But what do you mean by "feature selection code"? What features do you want? Or do you consider the 100,000 filtered pixel values features? What kind of features do you want to extract from each row of M (each set of 40 images)? There is no general purpose "feature detection code" that works for all images ever created in the history of the universe. You have to specify what algorithm you want for your particular images..

And you say "where the input matrix is the previous M matrix". What is the previous M matrix? At this point all we know of is one M matrix. Where did the other ones come from?

[Alyaa Sabra]

firstly thanks for your interest and your answer "Image Analyst" M matrix size 1725 row *100000 column i want perform any type of feature selection algrothem for face recognition in M matrix that contain features of 1725 images the goal of feature selection is minimize the number of features for example there are popurar type of feature selection named sequential feature selection i am sorry for my bad English language it is one M matrix , there isn't other matrix

[Image Analyst]

See http://www.mathworks.com/help/vision/examples/face-detection-and-tracking-using-camshift.html?prodcode=VP&language=en