Face detection and face recognition

I was always fascinated by face detection. Thanks to all the detective serials that I used to watch where cops compared faces of fugitives with their existing database! Sounded far-fetched then.

But face detection technology is here to stay.

What is face detection?

It is a technology that determines the exact locations and sizes of human faces in arbitrary (digital) images. It detects facial features and ignores anything else, such as buildings, trees and bodies.

Mostly face-detection algorithms focus on the detection of frontal human faces, whereas newer algorithms attempt to solve the more general and difficult problem of multi-view face detection. That is, the detection of faces that are either rotated along the axis from the face to the observer (in-plane rotation), or rotated along the vertical or left-right axis (out-of-plane rotation), or both.

The method devised by Viola and Jones, for example, uses Haar-like features. They used AdaBoost to train a classifier, which allows for a feature selection. The final classifier only uses a few hundred Haar-like features.

Needless to say that Aftek has worked on face detection and has achieved a very good hit rate with this, with a relatively low false detection rate.

Face recognition:

Face recognition is the next step to face detection. It is one of the most successful applications of image analysis and understanding. A general statement of face recognition problem can be founded as;

Given still or video images of a scene, identify or verify one or more persons in the scene using a stored database of faces. This requires an accurate detection of human faces in arbitrary scenes. When faces are located exactly in any scene, the recognition step can be used to identify or verify the detected faces.

Algorithms like PCA (Principle component analysis) and DCT (Discrete Cosine transform) are used for this.

What has Aftek done?
We have already implemented this solution in our ‘SpyGuard’, security product. AVI clip is captured from the motion sensitive camera and face (if present) is detected in the scene. If the face gets detected a notification is sent via a local alarm or an SMS to the home owner.

Face Recognition for a Bank:
Face recognition can be used to overcome the current security breaches in banks. We know that safe vaults in banks have limited access and only authorized persons can enter and access the lockers. The current methods are not foolproof. Here, the face recognition system can detect if the person entering the vault is authorized or not. We can do this by first capturing the images of the person and then verifying them with a database of available images.

Now to have the database in place, the solution can be trained with a snap of the bank locker holder while registering. And whenever the person wishes to access his/her locker, a snapshot will be captured by the camera (mounted on the wall) and compared with the earlier trained snap. Access to the facility will be given accordingly.

Kavita Chate (kc@aftek.com)
http://www.aftek.com/
http://spyguard.aftek.com/