摘要
Face recognition, a biometric method of identifying individuals using facial features, has attracted increasing interest and research over the last decade. In this paper, we propose a hierarchical scheme for face recognition. The proposed scheme consists of chin outline classification and holistic facial feature identification. Chin-shape information is characterized by chin curvature, the length of the face, and the ratio of face width to face length, all of which are scale-independent. After the holistic facial features were extracted based on Gabor faces, two-dimensional principal component analysis was used to condense the features. A series of experiments was conducted to assess system performance. The results confirm that our system can accurately recognize faces by the hierarchical integration of the feature-based information and the holistic one.
- 出版日期2010-12