Face identification based on TV videos

Abstract

Face recognition has been a main object of study, producing robust systems with high reliability. However, these systems face the inconvenience of long processing times which are, in many cases, unwanted. This paper presents a simple face recognition approach from TV video files based on the redundancy of information and a face recognition system used in regular face pictures. More precisely, the model relies on the wavelet transform for parameterize faces and support vector machines in order to classify them. After some experiments, the model has shown perfect accuracy in subject detection and recognition, with maximum delay of around 2 seconds.