| 摘要: |
| 针对Hough圆变换、最小二乘法椭圆拟合瞳孔中心定位算法中运算量大的问题，提出一种基于椭圆外切矩形的快速瞳孔中心定位算法。该算法从人眼图像中通过阈值分割得到瞳孔区域，对瞳孔区域进行边缘像素点的提取，然后利用椭圆外切矩形模型定位出瞳孔中心。通过Matlab对中国科学院自动化研究所公开的虹膜数据库中756张人眼图像进行遍历实验验证，所提算法定位准确率为98.3%，平均用时0.192 s。在同等条件下，与Hough圆变换、最小二乘法椭圆拟合算法进行对比，在保证定位准确率的同时，所提算法平均用时更少。实验结果表明，所提算法能够快速准确地完成瞳孔的中心定位。 |
| 关键词: 人眼图像 瞳孔中心定位 椭圆外切矩形 阈值分割 边缘像素点 |
| DOI： |
|
| 基金项目:国家自然科学基金资助项目（61674100,61774101） |
|
| A Fast Pupil Center Localization Algorithm for Near-eye Applications |
| ZHAO Haoran,JI Yuan,MU Tingzhou,CHEN Wendong,GAO Qin,RAN Feng |
| (Microelectronic Research and Development Center,Shanghai University,Shanghai 200444,China;a.Microelectronic Research and Development Center;b.School of Mechatronic Engineering and Automation,Shanghai University,Shanghai 200444,China) |
| Abstract: |
| To solve the problem of large amount of computation in the pupil center localization algorithm for Hough circle transformation and least squares ellipse method,this paper proposes a fast pupil center localization algorithm.The algorithm obtains the pupil region from the human eye image through threshold segmentation,extracts the edge pixel points from the pupil region,and then uses the elliptical circumscribed rectangle model to locate the pupil center.The traversal experiments of 756 human eye images in the iris database published by the Institute of Automation,Chinese Academy of Sciences are performed by Matlab.The accuracy of the proposed algorithm is 98.3%,and the average time is 0.192 s.Under the same conditions,compared with Hough circle transformation and least square ellipse fitting algorithm,the proposed algorithm takes less time on average while ensuring the location accuracy.The experimental results show that the proposed algorithm can locate the pupil center quickly and accurately. |
| Key words: human eye image pupil center localization elliptical circumscribed rectangle threshold segmentation edge pixels points |
《电讯技术》编辑部 