Two-Stage Data Reduction for a SVM Classifier in a Face Recognition Algorithm Based on the Active Shape Model
[ 1 ] Instytut Automatyki i Inżynierii Informatycznej, Wydział Elektryczny, Politechnika Poznańska | [ P ] pracownik
2011
rozdział w monografii naukowej
angielski
EN In this paper, two stage data reduction method for face identification with use of Support Vector Machine (SVM) classifier is evaluated. SVM Classification was performed for data acquired from contour description of 2200 faces of 100 persons. Face contours were extracted from frontal face images with use of Active Shape Model (ASM) method. Two stage PCA+LDA data reduction performance is measured in comparison with single stage PCA or LDA reductions. We propose to replace first stage PCA reduction with much simpler and less computationally intensive contour decimation.
647 - 656