Optimizing your database for accurate Face recognition

edit

What are the steps do we need to implement to preprocess the images in our database for a more accurate face recognition? I want to implement the code in C++