Evaluation of Local Tetra Pattern Features For Face Recognition

This paper presents an evaluation of two dense descriptors that have been or can be used in face recognition. Local Binary Pattern (LBP) is an established descriptor with many researchers focusing on its utility in Face Recognition. Local Tetra Patterns (LTrP) have been recently introduced and this paper explores the possibility of its utility in Face Recognition. A comparative analysis on the use of both these descriptors in face recognition has been carried out. Results have been presented on well-known ORL, Yale and FERET databases. It is worth stating that in the present experimental setup, LTrP is able to improve the recognition rate only marginally when compared to LBP.