Recognition of face orientation by divided hausdorff distance

In this paper we present a face orientation angle recognition system based on a modified Hausdorff distance. The Hausdorff distance determines the similarity between a given test image and a template. However, this operation is slow due to that the whole image is considered while using the original Hausdorff distance for matching. To address this matter we propose a divided Hausdorff distance which uses sub-regions of images while matching. Our experiment results show that the proposed method has better face orientation recognition and a high processing speed compared to original one.