How a Face Recognition can be achieved with a single image per person from the database if a query image is present

Question

I am developing Automatic Human Face Recognition Module. Whatever techniques present eg Eigen Face , Fisher Face , LBPH , PCA uses at-least 2 face images per person in the training database and this will be used to recognize an unknown query face image whether it is present in the database or not.

But My challenge is to develop a Face Recognition Module where only 1 face image is available per person in the database and I have to produce the output whether the unknown query image present in the database or not?

Is there any technique(s) please let me know?

Thank you in advance

Answer 1

I suppose two photos are required to develop a 3D image? Hence better comparison.

Otherwise, do what you can with a single photo:

• calculate SIFT, SURF, ORB or whatever descriptors for the reference photo,
• do the same for the queried photo (assumption: it may be slightly changed),
• use RANSAC, ICP or whatever iterative approach to approximately match descriptors from reference photo, to the query photo,
• if you've found a satisfying match, return true, otherwise - it may not be in the database, or the method is flawed, or the data is insufficient.