Author(s)

N. Sarris, G. Tzanetos and M.G. Strintzis

Abstract

A novel neural network technique for the detection of human faces in visual scenes Nikolaos Sarris, Georgios Tzanetos and Michael G. Strintzis Information Processing Laboratory, Electrical and Computer Engineering Department, Aristotle University ofThessaloniki, Greece Abstract This paper presents a novel neural network technique for the detection of human faces and their basic characteristics (eyes and mouth) in complex visual scenes. The proposed method utilises a two dimensional back-propagation neural network which is structured in such a way as to relate observations in local overlapping two-dimensional areas of the image scene. Results have been found to be satisfactory even in scenes where the face is slightly rotated, as long as both eyes are still visible. 1. Introduction We present a novel neural network method for the detection of human faces in image scenes. The proposed technique examines the image as a vector of its pixels arranged in lexicographic ord

Keywords