We propose a novel face detection algorithm in order to improve higher detection rate of face-detector than conventional haar - adaboost face detector. Our purposed method not only improves detection rate of a face but decreases the number of false-positive component. In order to get improved detection ability, we merged two classifiers: adaboost and support vector machine. Because SVM and Adaboost use different feature, they are complementary each other. We can get 2~4% improved performance using proposed method than previous our detector that is not applied proposed method. This method makes improved detector that shows better performance without algorithm replacement.

Date Published: 22 August 2014
PDF: 6 pages
Proc. SPIE 9286, Second International Conference on Applications of Optics and Photonics, 92864W (22 August 2014); doi: 10.1117/12.2064815

Published in SPIE Proceedings Vol. 9286:
Second International Conference on Applications of Optics and Photonics
Manuel Filipe P. C. Martins Costa; Rogério Nunes Nogueira, Editor(s)