Redundant Class-Dependence Feature Analysis Based on Correlation Filters Using FRGC2.0 Data

Abstract

In this paper we propose a new method called redundant class-dependence feature analysis (CFA) based on the advanced correlation filters to perform robust face recognition on the Face Recognition Grand Challenge (FRGC) data set. The FRGC contains a large corpus of data and a set of challenge problems. The data is divided into training and validation partitions, with the standard still-image training partition consisting of 12,800 images, and the validation partition consisting of 16,028 controlled still images, 8,014 uncontrolled stills, and 4,007 3D scans. We have tested the proposed CFA method and compared it with the PCA and LDA methods in a recognition scenario on the FRGC2.0 data. The preliminary results show that the CFA outperforms the other two compared methods in our experiments. We also show the improved performance of the CFA method on the FRGC experiments #1 and #4.