%0 Conference Proceedings
%4 sid.inpe.br/sibgrapi@80/2007/08.02.10.54
%2 sid.inpe.br/sibgrapi@80/2007/08.02.10.54.45
%@doi 10.1109/SIBGRAPI.2007.44
%T A learning-based eye detector coupled with eye candidate filtering and PCA features
%D 2007
%A Leite, Bruno de Brito,
%A Pereira, Eanes Torres,
%A Gomes, Herman Martins,
%A Veloso, Luciana Ribeiro,
%A Santos, C´ı,
%A Nascimento, cero Einstein do,
%A de Carvalho, João Marques,
%@affiliation Departamento de Sistemas e Computação, Universidade Federal de Campina Grande
%@affiliation Departamento de Sistemas e Computação, Universidade Federal de Campina Grande
%@affiliation Departamento de Sistemas e Computação, Universidade Federal de Campina Grande
%@affiliation Departamento de Engenharia Elétrica, Universidade Federal de Campina Grande
%@affiliation Departamento de Engenharia Elétrica, Universidade Federal de Campina Grande
%@affiliation Departamento de Engenharia Elétrica, Universidade Federal de Campina Grande
%E Falcão, Alexandre Xavier,
%E Lopes, Hélio Côrtes Vieira,
%B Brazilian Symposium on Computer Graphics and Image Processing, 20 (SIBGRAPI)
%C Belo Horizonte, MG, Brazil
%8 7-10 Oct. 2007
%I IEEE Computer Society
%J Los Alamitos
%S Proceedings
%K eye detection, neural networks, principal component analysis, integral image.
%X In this work, we present a system based on a Neural Network classifier for eye detection in human face images. This classifier works on eye candidate regions extracted from a face image and represented by a reduced number of features, selected by Principal Component Analysis. The regions are determined considering that in an image window containing the eye, the gray level distribution will generally assume a pattern of adjacent light-dark-light horizontal and vertical stripes, corresponding to the eyelid, pupil and eyelid, respectively. For training, validation and testing, a database was built with a total of 4,400 images. Experimental results have shown that the proposed approach correctly detects more eyes than any of two existing systems (Rowley-Baluja-Kanade and Machine Perception Toolbox), for eye location error tolerances from 0 to 5 pixels. Considering an error tolerance of 9 pixels, the correct detection rate achieved was above 90%. .
%@language en
%3 gomes-EyeDetection.pdf