积分投影与连通域法结合的人眼定位

Eyes Localization Based on Integral Projection and Connected Domain Algorithm

人脸识别是生物识别的重要组成部分,人眼定位是人脸识别过程中最关键的环节。人眼定位中,最难的就是区分眼睛和眉毛。论文先用Adaboost法粗定位人脸,获取人脸RIO区域,再利用积分投影法初步估计眼睛和眉毛的区域,结合人脸"三庭五眼"的特征,单独分割出眼睛和眉毛区域,对眉眼区域形态学处理,最后采用连通区域法精确定位眼睛。在ORL人脸库上检测,该方法获得了90.2%的准确率。实验结果表明该方法不仅能准确区分眼睛和眉毛,而且还能快速地定位人眼的位置。

Face recognition is an important part of biological recognition.Eye location is the key link in the process of face recognition.And the most difficult part of eye positioning is to distinguish between eyes and eyebrows.In this paper,the Adaboost method is used to locate face roughly and getting the RIO area firstly.Then,the Integral projection method is used to estimate the area of the eyes and eyebrows.The eye and eyebrow area are cut out combining with the features of"three chambers and five eyes"of human face and Morphological processing the eye and eyebrow area.Finally the connected region is used to locate eyes accurately.The method is tested on ORL face database and achieves a 90.2%accuracy rate.Experimental results show that this method can not only distinguish eyes and eyebrows and also can locate human eyes quickly.