Eyes detection combined feature extraction and mouth information

Abstract

In this paper, we propose an eye localization approach combining facial feature extraction and mouth information in color image. This approach includes feature extraction, face mask label, mouth location and eye region detection. In order to quickly obtain facial information, the coarse facial region is first localized by V-J filter, and then the compact facial region by our proposed filter can be decide and non-face region will be filtered out. Next, we label a face mask to modify skin region, extract the facial features and locate mouth information. Finally, the position of two eyes is detected based on the prior information. We present the experimental results on the various cases, including profile face, frontal face, and whether glasses or not, to demonstrate the effectiveness of our method. © 2013 Springer International Publishing.