Бумбаров, Огнян; Bumbarov, Ognyan; Соколов, Страхил; Sokolov, Strahil; Момчеджиков, Михаил Благоев рец.; Momchedzhikov, Mihail Blagoev rew.
Разпознаване на изображенията на лица с използване на анализ по главни компоненти
Face recognition using principal component analysis
PROCEEDINGS OF THE TECHNICAL UNIVERSITY - SOFIA, 55, 2005 - Anniversary “Sixty Years Technical University - Sofia”, pp. 215-224
TU - Sofia
изображения - математическо моделиране
A face recognition algorithm is presented – for fontal face images. It is based on skin color detection and principal component analysis. Our algorithm includes: face detection based on human skin modeling, Principal Component Analysis (PCA) and classification. PCA is a statistical technique for dimensionality reduction of data. PCA on a set of n image faces can produce up to n components, each accounting for a decreasing amount of variance. A face is presented as a vector in a space defined by these components (a “face space”). An average face of the set is obtained. Later on, each face of the set is subtracted from the average face. The results of the subtraction (images of difference faces) are transformed into their “eigenfaces” (projected into “face space”). A new image face for identification is transformed into its “eigenface”. Once the eigenfaces are created, identification becomes a pattern recognition task.
Рец. проф. д-р Михаил Момчеджиков. - С ил., формули