Extended JSSL for Multi-Feature Face Recognition via Intra-Class Variant Dictionary

This paper focuses on how to represent the testing face images for multi-feature recognition. The choice of feature is critical different features sample contribute differently joint similar and specific learning (JSSL) has been effectively applied in In JSSL, although representation coefficient divided into coefficient, there disadvantage that training cannot well, because are probable expressions, illuminations disguises images. We think intra-class variations one person can be linearly represented by those other people. order solve well paper, we extend JSSL propose extended (EJSSL) EJSSL constructs variant dictionary variation between uses proposed method perfectly experimented some available databases, its performance superior many current recognition methods.