Pose and Expression Normalization for Face Recognition Using Delaunay Triangulation

DOI

10.2991/aer.k.201221.057How to use a DOI?

Keywords

Face Recognition, Expression Normalization, Pose Normalization, Delaunay Triangulation, Neutral Face Reference

Abstract

Detecting face image with various poses and expressions is a challenging problem especially if the face database consists of only neutral faces. To solve that problem, the pose and expression of face image should be neutralized first. This study proposes a method to recognize the face image by making the input image to be neutral in pose and expression to make it similar to the face images in the virtual database. The frontal face image needs to be neutralized for the expression using Delaunay Triangulation face model. For correcting pose view, a mirroring method is included in the transformation process. The final image that has obtained is evaluated by comparing it to the virtual face database. The face database; namely; Tarrlab face database is used for the experiment. Local Binary Pattern is chosen for the face recognition method. On average, this proposed study has been achieved improvement in recognition accuracy by 47.34%. A significant improvement can be shown at input images that have the extreme angle pose like 30 and 45 degrees.

Copyright

© 2020, the Authors. Published by Atlantis Press.

Open Access

This is an open access article distributed under the CC BY-NC license (http://creativecommons.org/licenses/by-nc/4.0/).

Download article (PDF)

TY  - CONF
AU  - Muhammad Yusuf Fadhlan
AU  - Dianthika Puteri Andini
PY  - 2020
DA  - 2020/12/22
TI  - Pose and Expression Normalization for Face Recognition Using Delaunay Triangulation
BT  - Proceedings of the International Seminar of Science and Applied Technology (ISSAT 2020)
PB  - Atlantis Press
SP  - 344
EP  - 350
SN  - 2352-5401
UR  - https://doi.org/10.2991/aer.k.201221.057
DO  - 10.2991/aer.k.201221.057
ID  - Fadhlan2020
ER  -