Importance

Loading Importance Indicators...

Abstract

A face classification system which receives video data representative of a scene and extracts therefrom data representative of a face which is present in the scene. The facial video data is applied to a homomorphic filter to eliminate the effects of lighting changes in the scene. The positions of the eyes and mouth are located to form two sub-divisions, one in relation to a first line joining the eyes and the other in relation to a second line perpendicular to the first line and passing through the nose. A feature vector which is representative of the face and is rotation, scaling, translation and grey level intensity invariant is then produced by applying a recursive second order sub-division of moments to the filtered data. Such sub-division of moments is constrained to act first on the sub-divisions formed by the aforesaid lines, thereby reducing the effect of noise. The feature vector may be applied to a classifier which employs multi-layer perception to compare it with stored feature vectors of faces in one or more previous scenes to determine if the faces match. The system may be used, for example, for allowing access to secure areas or to secure computer work stations.

Loading the Abstract Image...

First Claim