A robust eye detection method based on multi-neighborhood block with weight and SVM on gray image
Liming Yan, Xinzhu Sang · Proceedings of SPIE, the International Society for Optical Engineering/Proceedings of SPIE · 2013
Eye detection plays a vital role in intelligent recognition system. In this paper, a novel of robust and fast eye detection algorithm is proposed. The process of eye detection can be divided into two stages. Firstly, coarse eye candidates are extracted based on multi-neighborhood blocks with weight; Secondly, eye candidates validation is testified on support vector machine (SVM).Data from different face databases such as self-built, JAFFE, FERET are used to demonstrate the effectiveness and robustness of the proposed method.