2024年2月25日发(作者:索蕙兰)
基于 HEIV 算法改进的瞳孔中心定位
贾彩琴;杨剑
【期刊名称】《计算机工程与设计》
【年(卷),期】2015(000)012
【摘 要】针对瞳孔中心定位的精确度问题,采用基于 AdaBoost 的人脸检测算法确定人脸区域与初始眼睛位置,由于人眼轮廓具有的特征,以椭圆为模型,用
Sobel 算子边缘检测眼睛区域,进行椭圆拟合来获取眼睛的轮廓线,将椭圆拟合的中心点作为瞳孔的精确位置。为提高定位的精确度,提出异方差(heteroscedastic errors-in-variables,HEIV)方法来实现瞳孔中心定位,较最小二乘法更加精确。实验结果表明,在正面人脸的情况下,异方差椭圆拟合算法与最小二乘拟合算法相比,计算量小,可精确定位瞳孔中心,平均误差比最小二乘法提高了89.1%,该方法在速度和准确性方面均具有良好的性能。%Aiming at
the problem of the precision of localization of pupil center,the face area
and the initial eye position were de-termined using face detection
algorithm based on AdaBoost,because of the features of the outline of
human eyes,the pupil was modeled as an ellipse,the eye regions were
detected and obtained using edge of Sobel operator,and the eye ellipse
contour line was got through ellipse center point of ellipse
fitting was taken as the precise location of the improve the
positioning accuracy,heteroscedastic (heteroscedastic errors-in-variables,HEIV)method was used to realize the localization of pupil
center,which was more accurate than the least square
mental results show that,in the situation of the front face,the
heteroscedasticity ellipse fitting algorithm,compared with the least square
fitting algorithm,has small amount of calculation,and it can more precisely
localize the pupil ed with the least square method,the
average error is im-proved by 89.1%.The proposed method has good
performance on both speed and accuracy.
【总页数】6页(P3273-3278)
【作 者】贾彩琴;杨剑
【作者单位】中北大学 计算机与控制工程学院,山西 太原 030051;中北大学 计算机与控制工程学院,山西 太原 030051
【正文语种】中 文
【中图分类】TP391.4
【相关文献】
1.基于椭圆拟合的瞳孔中心精确定位算法研究 [J], 余罗;刘洪英;皮喜田;陈为民;许帅;严旭
2.单分子中心定位精度与信噪比关系及中心定位算法改进研究 [J], 翟人宽;李小卫;孙洁林;邵志峰
3.基于视频眼震图的瞳孔中心定位方法研究 [J], 王长元;史学颖
4.基于改进的SIFT算子和SVM分类器的瞳孔中心定位 [J], 田耘;甄雯;赵海军
5.基于星射线法的瞳孔中心定位 [J], 马启玮;韩慧妍;韩燮;张增一
因版权原因,仅展示原文概要,查看原文内容请购买
2024年2月25日发(作者:索蕙兰)
基于 HEIV 算法改进的瞳孔中心定位
贾彩琴;杨剑
【期刊名称】《计算机工程与设计》
【年(卷),期】2015(000)012
【摘 要】针对瞳孔中心定位的精确度问题,采用基于 AdaBoost 的人脸检测算法确定人脸区域与初始眼睛位置,由于人眼轮廓具有的特征,以椭圆为模型,用
Sobel 算子边缘检测眼睛区域,进行椭圆拟合来获取眼睛的轮廓线,将椭圆拟合的中心点作为瞳孔的精确位置。为提高定位的精确度,提出异方差(heteroscedastic errors-in-variables,HEIV)方法来实现瞳孔中心定位,较最小二乘法更加精确。实验结果表明,在正面人脸的情况下,异方差椭圆拟合算法与最小二乘拟合算法相比,计算量小,可精确定位瞳孔中心,平均误差比最小二乘法提高了89.1%,该方法在速度和准确性方面均具有良好的性能。%Aiming at
the problem of the precision of localization of pupil center,the face area
and the initial eye position were de-termined using face detection
algorithm based on AdaBoost,because of the features of the outline of
human eyes,the pupil was modeled as an ellipse,the eye regions were
detected and obtained using edge of Sobel operator,and the eye ellipse
contour line was got through ellipse center point of ellipse
fitting was taken as the precise location of the improve the
positioning accuracy,heteroscedastic (heteroscedastic errors-in-variables,HEIV)method was used to realize the localization of pupil
center,which was more accurate than the least square
mental results show that,in the situation of the front face,the
heteroscedasticity ellipse fitting algorithm,compared with the least square
fitting algorithm,has small amount of calculation,and it can more precisely
localize the pupil ed with the least square method,the
average error is im-proved by 89.1%.The proposed method has good
performance on both speed and accuracy.
【总页数】6页(P3273-3278)
【作 者】贾彩琴;杨剑
【作者单位】中北大学 计算机与控制工程学院,山西 太原 030051;中北大学 计算机与控制工程学院,山西 太原 030051
【正文语种】中 文
【中图分类】TP391.4
【相关文献】
1.基于椭圆拟合的瞳孔中心精确定位算法研究 [J], 余罗;刘洪英;皮喜田;陈为民;许帅;严旭
2.单分子中心定位精度与信噪比关系及中心定位算法改进研究 [J], 翟人宽;李小卫;孙洁林;邵志峰
3.基于视频眼震图的瞳孔中心定位方法研究 [J], 王长元;史学颖
4.基于改进的SIFT算子和SVM分类器的瞳孔中心定位 [J], 田耘;甄雯;赵海军
5.基于星射线法的瞳孔中心定位 [J], 马启玮;韩慧妍;韩燮;张增一
因版权原因,仅展示原文概要,查看原文内容请购买