Image Searching with Eigen Faces and Facial Characteristics

Abstract

Automated face recognition is an application domain for automatically identifying or verifying a person in a digital image or a video frame from a video source. This is becoming an increasingly important function in many areas ranging from static matching of controlled photographs as in mug shots matching and credit card verification to surveillance video images. However for many years face recognition technology has been focused on one face: frontal, well illuminated and looking directly at the capture device. In reality the situation is different. Human face is being viewed in variety of poses, under a range of illumination conditions and changing expression in real world. In recent years, a considerable number of researches have been carried out to deal with each of these forms of variations individually. However combining techniques to deal with multiple forms of variation at once has received less scrutiny. In this research we utilize a number of established techniques which cater for these forms of variations individually and unify them into a single novel algorithm for better recognition of human faces. As part of this project the application Face Reco is implemented using C#.net and it can be used to automatically recognize faces in both still and video sequences.