A statistic-based approach for automatic multi-view face detection and pose estimation

Multi-view face detection has become one of the most attractive research topics in the field of computer vision. In this paper, a novel statistic-based system for automatic multi-view face detection and pose estimation is proposed. Our approach constructs a multi-level framework utilizing multiple appearance-based learning methods to build corresponding face detectors and pose estimators, and hierarchically filters human faces. Contributions include the coarse-to-fine structure considering both efficiency and accuracy, different facial features representing low- and high-dimensional information, and statistic discriminant function regularizing divergent features. The results not only demonstrate the superiority of automatically identifying facial images, but also verify the ability in estimating various poses.