Robust Face Expression Recognition with OSCAR

Abstract

In this paper, a novel classifier is used. Our method uses Principal Component Analysis (PCA) as expression features representation and employs octagonal shrinkage and clustering algorithm for regression (OSCAR) as expression classifier. PCA can reduce the dimensions of feature vectors; and OSCAR can classify expression accurately. Experiments of human are performed on the JAFFE database, and compared to the other classifiers, our method can get better recognition ratio.