Face Detection with Haar Cascades Method

Abstract

Face detection is a process of finding and identifying human faces in digital images. It is widely applied in social media, digital services, security systems, robotic and other applications. A human frontal up-right face detector had constructed in this paper by Haar Cascade method using MATLAB. Input images are often scaled to obtain a variety of different size faces, rather than scaling the detection window. The integral image method is also used to reduce the computation time of features value. The relationship between the scale factor with accuracy and computational time had been studied in this paper. In overall, the suggested scale factor in this paper is 1.1 to achieve higher accuracy and shorter computation time. To enhance the face detector created, extended rotated Haar-like features by Messom and Barezak in 2006 can be considered to detect rotated human faces.