Detecting Human Heads and Face Orientations Under Dynamic Environment

We propose a two-step method for detecting human heads and estimating face orientations under the dynamic environment. In the first step, the method employs an ellipse as the contour model of humanhead appearances to deal with wide variety of appearances. Our method then evaluates the ellipse to detect possible human heads. In the second step, on the other hand, our method focuses on features, such as eyes, the mouth or cheeks, inside the ellipse to model facial components. The method evaluates not only such components themselves but also their geometric configuration to eliminate false positives in the first step and, at the same time, to estimate face orientations. In the both steps, our method employs robust image-features against lighting conditions in evaluating the models. Consequently, our method can correctly and stably detect human heads and estimate face orientations even under the dynamic environment. Our intensive experiments show the effectiveness of the proposed method.