Frontal face localization using linear discriminant
Lingmin Meng, T.Q. Nguyen · 2003
This paper introduces a novel face detection approach using linear discriminant to detect frontal faces in grayscale images. By modelling both faces and clutters as Gaussian distributions, an optimal discriminant template is developed to minimize the detection Bayesian error. Our simulation indicates that the proposed approach outperforms conventional template-based methods such as matched filter and eigenface methods.