Representation of Pose Invariant Face Images Using SIFT Descriptors

The choice of a face database should solemnly depend on the problem to be solved. In this research work, we use the Face Recognition Technology (FERET) database to address the challenge of face pose variations. The Scale Invariant Feature Transform (SIFT) is used to represent these face images in the database. SIFT has been proven to be a robust and a powerful method for general object detection in the past years. This method is now popular in the field of face recognition for purposes of extracting key points which are scale and orientation invariant from the face image. This work demonstrates that through extracting SIFT features from different face image patches and at different sigma σ values, a face pose can be classified towards better pose invariant face recognition.