Fast face detection using boosted eigenfaces
Anand Mohan, N. Sudha · 2009
This paper describes a new eigenface based face detection using boosted eigen features. Eigenfaces have long been used for face detection and recognition. The basic detection and recognition system works by projecting the face images onto a feature space that spans significant variations among the training set. But the distance from the face space is not a reliable measure to classify faces from non-faces as some of the non-faces may also lie close to the face space. We propose to build a better classifier by boosting a set of weak classifiers built from the projections onto the eigen vectors of the face space. The proposed system provides significantly better performance compared to the distance measure. Also, we propose to improve the speed of detection in real images using FFT.