作者: K. Sobottka , I. Pitas
关键词: Face (geometry) 、 Computer vision 、 Computer science 、 Facial recognition system 、 Artificial intelligence 、 Segmentation 、 Active contour model 、 Face detection 、 Image segmentation 、 Pattern recognition 、 Hue 、 Feature extraction
摘要: The authors present a new approach for automatically segmentation and tracking of faces in color images. Segmentation is performed by evaluating shape information. First, skin-like regions are determined based on the attributes hue saturation. Then with elliptical selected as face hypotheses. They verified searching facial features their interior. After reliably detected it tracked over time. Tracking realized using an active contour model. exterior forces snake defined features. push or pull snaxels perpendicular to snake. Results shown image sequence consisting 150 frames.