First define the parameters of the attractiveness, then select a suitable software/tool of image processing that deals all these parameters, then put a standard measure of the parameters of attractiveness for that software/tool. For that you can contact any computer scientist/Mathematician who deal with image processing......

Dear Subrata, easier said than done ;)

yeah i know its not so easy. but first  which factor of a face you think to be part of attractiveness?? Have you include complexion, shape and other factors of attractiveness? Attractiveness is a relative mater, one may seem to be attractive to some one and for some other one not. So its too hard to make a general model. But may be done with some absolute type of factors of face, absolute means almost similar to everyone.

Subrata, do you think that it might be possible to teach a neural network with examples of faces differing in attractiveness so that it might figure out how to categorize future inputs? There are such databases available.

Well,

If #s are beautiful then all faces(as an image) are attractive else subjectivity factors in !!!

Neural response of people (who judge- on some scale) has to considered for decision making as you build the network ...