Face Recognition Using A New Class-dependent Feature Analysis Method
Ding Xiaoqing · Video Engineering · 2007
In this paper,the author describes his research on Class-dependent Feature Analysis Method using correlation filters to extract features in human face recognition.The author adopts the Minimum correlation energy rule to correlation filters to im- prove the recognition capability of correlation filters.The Verification Rate at 0.1% of FAR rises from 90% to 94.46% on gray scale.The new algorithm also saves more training time.