一种基于区域投影的人眼精确定位方法

A precise eye localization method based on region projection

针对传统灰度投影方法抗干扰能力较差的弱点,提出了一种基于区域投影的人眼精确定位方法。考虑到投影过程中的二维特性,在水平和垂直方向将眼睛图像分成不相重叠的区域,分别将各区域内的灰度值投影获得瞳孔的候选区域并将其扩展获得瞳孔窗口,利用灰度特性通过边界跟踪的方法实现了对瞳孔中心的精确定位,给出了人眼定位精确度判定准则,采用Caltech faces和JAFFE数据库进行测试。实验表明,与传统的投影方法比较,本文方法具有较强的鲁棒性,人眼定位精度更高。

Due to the poor anti-jamming ability of traditional gray projection method,we propose a new method based on region projection for precise eye localization.Firstly,taking the two-dimensional characteristics into account in the process of projection,the eye image was divided into several non-overlapping regions,and the candidate region of pupil was obtained by horizontal projection and vertical projection.The region was expanded and a pupil window was got.Then,the pupil center was localized exactly by border tracking with the gray characteristics.Finally,the criterion to evaluate the localization accuracy of human eye was worked out,and the algorithm was tested in Caltech face databases and JAFFE database.Experimental show that the method is effective and robust,and has higher eye-localization precision than traditional projection methods.