Example-Based Learning for Face Image Recognition

Appearance-based face recognition was extended into across poses scenarios in this paper. Information related with pose rotation was discarded in traditional multi-view works but was learned from sample face images in present. Face recognition was implemented by reconstructing frontal view features using transformation matrix. Experiments proved that the proposed method was a much effective approach for recognition across poses. Compared with model-based approaches in face recognition, the proposed work is not dependent on heavy computation and has merit of easy implementing in real conditions. Popular feature extractions, least square (LS) and total least square (TLS) solution were compared as well.