3D Face Recognition Using Nose Lines

An approach to 3D face recognition using BJUT-3D database and FRAV3D database has been presented in this paper. The proposed algorithm uses central part of a face for extracting facial features. Therefore, other parts like chin, hairs and neck that don t contain usable information are eliminated. In this method, by focusing on the centers of 3D faces and extracting diagonal lines of nose area, required features for comparing faces will be available. The results prove that diagonal lines are informative enough to be used in face recognition applications. Diagonal lines on nose region offer 100% performance using neutral images of BJUT-3D database. In other experiments on FRAV3D database, we obtain the highest performance 85% using weighting methods, more nose area samples and images with expressions. Acquired results demonstrate the accuracy and efficiency of our algorithm which could be a promising method.

کلیدواژه ها:

3D face recognition ، correlation coefficient ، expression variation ، nose diagonal lines