extract feature vector matrix for neural network

hello i want to extract SIFT features from each face for face recognition using neural networks. When i am running the code given at official SIFT website :

         [image, descriptors, locs] = sift('1.pgm');

I am getting three matrices

             image 58x128 double
             descriptors 112x92 unit8
             locs 58x4 double

What should I choose as a feature vector for my neural network ?

[Greg Heath]

If you have 58 images what do the numbers

 112 and 92 represent ???

Greg

[Newman]

Dear sir I don't have 58 images but 58 keypoints per image.

The image is just one which in this case is '1.pgm' and it has found 58 keypoints in it.

It returns an image which is nothing bit the double of the input image. The descriptor is just a 112 dimensional vector of the 92 descriptions and locs represents their location in double format .

Everything is for one image .