ASF-LLRDA: Locality-regularized Linear Regression Discriminant Analysis with Approximately Symmetrical Face Preprocessing for Face Recognition

Abstract

Face recognition is a crucial task in numerous applications, but it has difficulties because of the high-dimensional nature of facial photos, limited sample sizes, variations in illumination, and facial expressions. This paper presents a novel face recognition approach to overcome these challenges by combining the advantages of the Approximately Symmetrical Face (ASF) preprocessing strategy and the Locality-regulated Linear Regression Discriminant Analysis (LLRDA) method. The proposed method, called ASF-LLRDA, makes use of ASF to generate axis-symmetric face images for reducing the impact of illumination and variations in facial expressions, followed by LLRDA to extract discriminative features and project the data into a lower dimensional space. Locality-regulated Linear Regression (LLRC) is further utilized as the classifier. Experimental results on the Yale-B face dataset demonstrated the superiority of the proposed method compared to the baselines.