Combing Haar and MBLBP features for face detection using multi-exit asymmetric boosting

This paper proposes a visual face detection framework that enables fast image processing while achieving high detection rates. The proposed framework combines both multi block local binary pattern (MBLBP) features and Haar- like features using multi-exit asymmetric boosting for robust face detection. In this framework, the integral image is utilized to facilitate rapid extraction for both the MBLBP features and the Haar-like features. Experimental results showed that combing MBLBP and Haar-like features can achieve better detection rate than each of them can do individually.