A real-time face detector using ellipse-like features

In this paper, we present a real-time face detector based on the detection of ellipse-like face shapes and face texture features. First, we propose a new ellipse-like detection algorithm based on improved fast Hough transform techniques. Then a face shape detector is developed that uses rules to detect the ellipse-like shape detector. Finally, we utilize a texture matching method through multiple stages of classifiers to determine whether the candidate region contains face. Our experiments show indeed that the proposed detector is accurate and efficient in real-time applications.