Frontal Face Detection Algorithms Based on Support Vector Machines and Maximum Likelihood

Face detection is a key problem in building automated systems that perform face recognition/ verification, modelbased image coding, face tracking, and surveillance. Two algorithms for face detection based on either support vector machines or maximum likelihood estimation are described and their performance is tested on a collection of single images from the M2VTS database that depict one frontal face in front of a uniform background using the false acceptance and false rejection rates as quantitative figures of merit. Moreover, we demonstrate how the maximum likelihood face detection performs, when single images that depict multiple frontal faces in front of an nonuniform background are processed.