A fast eye location method using ordinal features

Automatic facial feature location is an important step in humancomputer interaction and automatic face recognition. In this paper, a new eye location method is proposed. A classifier based on ordinal features is constructed by training on samples. It is used to evaluate the eye candidates discovered by the detector trained using AdaBoost. False positive responds are abandoned and eye centers are calculated by the evaluation result. The experimental results demonstrate that our method achieves good location performance. Keywords Eye location, Ordinal features

• Type: Conference
• Year: 2006
• Where: ACMACE
• Authors: Xuetao Feng, Yangsheng Wang, Li Bai
• Keywords: ACMACE 2006, Eye Location, Eye Location Method, Ordinal Features