Skip to Main Content
Requirement for a person to face a camera for face identification process may no longer be necessary if the face recognition system is robust against variation of facial pose. In this paper, we proposed a face recognition method which remains reliable even in very large head pose variations. In this method, feature from local regions of face are extracted after employing both discrete cosine transform and discrete wavelet transform. Learning strategy is then applied to infer the relationship between face in a given pose and its frontal view. Results we obtained are very promising considering that our proposed method solely relies on a single gallery image. We also demonstrated the high performance of our method in a condition whereby the face images are of low-resolution quality.