Two-Dimensional Robust Neighborhood Discriminant Embedding in face recognition

Abstract

This paper explores the use of Two-Dimensional Robust Neighborhood Discriminant Embedding (2D-RNDE) as a means to improve the performance and robustness of face recognition. 2D-RNDE is based on graph embedding framework and Fisher's criterion, it can utilize the original two-dimensional image data directly and takes into account the Individual Discriminative Factor (IDF) which is proposed to describe the microscopic discriminative property of each sample. The purpose of our algorithm is to gather the within-class samples closer and separate the between-class samples further in the projected feature subspace after the dimensionality reduction. Furthermore, another informative feature extraction method called circular pixel distribution (CPD) is proposed and applied to enhance the robustness of our algorithm. Experiments with the Olivetti Research Laboratory (ORL) face dataset are conducted to evaluate our method in terms of classification accuracy, efficiency and robustness.