I need some face similarity metric. I know about PCA,LDA approaches that uses raw pixels data, but they are not scale invariant(I know that I can resize image but I think that's not good solution). Maybe I need to use AAM,ASM aproaches? But I'm interested in metric that gives me the greatest separability of faces. At first I don't need automatic detection of metric features, for example imagine that I can mark all faces by hand.
Haven't used them first hand but I'm pretty sure that [SIFT] "Scale Invariant Feature Transform" is scale invariant. And I know that they are used, with great success, in face recognition.