How to use Hog featurs in face recognition?

Hi, Im trying to write program to face recognition using SVM and Hog featurs. I saw this code for adding Hog feature of image 'im' to the classifier database:

trainingFeatures(featureCount,:) = extractHOGFeatures(im);

Now the output of 'extractHOGFeatures' is a vector of 1 dimension, so how could SVM work with this kind of data? I d'ont know a lot on SVM, but as far as I know its work on 2-dimensional data to find a hyperline, so how its can work with 1-dim vector?

and another question, suppose I have few images of face that I need to recognized so I can train my classifier to find that face in a query image. How I can teach him to ignore other faces? I can set some threshold for acuuracy between the classifier and the query image?

thank you.