Face verification using Local Graph Stucture (LGS)

Abstract

Verification of human face plays an important role in social interaction, such as recognition and security. In this paper, a novel generative approach for face verification based on Local Graph Structure (LGS) description is proposed. A generic face model is considered as a collection of LGS-histograms. Local Graph Structure (LGS) operator is defined as a gray-scale invariant texture measure, derived from a general definition of texture in a local neighborhood. The histograms of the LGS were used for verification, various distance/similarity measures classifiers, namely, the Euclidian distance, the City block distance, the cosine similarity measure and the mahalanobis cosine. The experiments results on standard ORL face database images demonstrated the effectiveness of the proposed method.