Score Fusion of 3DLBP and ACDN+P for 3D Face Recognition

  • João Baptista Cardia Faculdade de Tecnologia "Dep. Júlio Julinho Marcondes de Moura" e Departamento de Computação, Universidade Federal de São Carlos - UFSCar - Brazil
  • Aparecido Nilceu Marana Universidade Estadual Paulista e Departamento de Computação, Universidade Federal de São Carlos - UFSCar - Brazil

Abstract

Many situations of our everyday life require our identification. Biometrics-based methods, besides allowing such identification, can help to prevent frauds. Among several biometrics features, face is one of the most popular due to its intrinsic and important properties, such as universality, acceptability, low
costs, and covert identification. On the other hand, the traditional automatic face recognition methods based on 2D features can not properly deal with some very frequent challenges, such as occlusion, illumination and pose variations. In this paper we propose a new method for face recognition based on the fusion of 3D low-level local features, ACDN+P and 3DLBP, using depth images captured by cheap Kinect V1 sensors. In order to improve the low quality of the point cloud provided by such devices, Symmetric Filling, Iterative Closest Point, and Savitzky-Golay Filter are used in the preprocessing stage of the proposed method. Experimental results obtained on EURECOM Kinect dataset showed that the proposed method can improve the face recognition rates.

Published
2016-08-05
How to Cite
CARDIA, João Baptista; MARANA, Aparecido Nilceu. Score Fusion of 3DLBP and ACDN+P for 3D Face Recognition. Journal on Advances in Theoretical and Applied Informatics, [S.l.], v. 2, n. 1, p. 34-39, aug. 2016. ISSN 2447-5033. Available at: <https://revista.univem.edu.br/jadi/article/view/1660>. Date accessed: 27 apr. 2024. doi: https://doi.org/10.26729/jadi.v2i1.1660.