asked 2013-05-21 14:19:24 -0600Adrianos01
My Project is to write a Program that recognize faces with Video Camera and compare the recognized Face if the recognized face in database. If it not in Database send Email.
I read in google the face detection and face recognition from OpenCV.
Face detection and Face recognition: learns a subspace for each person and when i input a new image it measures the distance to all subspaces it has previously learned and chooses the nearest one.
My Question is how can i sort and check that the image is too far from all learned subspaces. Also check the image is the same in the database:
I should realise that with OpenCV.
answered 2013-05-22 03:11:03 -0600berak
the desired threshold value depends on a number of factors:
the algorithm you choose. ( fisher/eigen/lbp FaceRecognizers all have differently sized feature-spaces and use different norms to evaluate the comparison )
your face-db. yes, you need to have it almost complete(say, full), to determine a value, that makes sense.
as for the last topic, how to find if an img is already in the db : this is much easier, it will come out with (almost) zero distance.
Asked: 2013-05-21 14:19:24 -0600
Seen: 342 times
Last updated: May 22 '13