A novel expanding sample method for single traning sample face recognition

Abstract

Single training sample face recognition(STSFR) is a tricky problem. It can cause a sharp decline in face recognition(FR) rate since each person only has one training sample. To solve this problem, a novel expanding sample method is proposed in this paper. Based on the fact that the intra-class facial variations can be shared across different persons, the intra-class facial variations are used to expand training sample. Firstly, the intra-class facial variations can be obtained from a face database. Secondly, a single training sample database is expanded by merging these intra-class facial variation images and single training samples in the single training sample database. Thirdly, the expanding database is used to perform experiment. Experiments on an expanding database demonstrate that a better performance is obtained by using the expanding sample method.