首页> 外国专利> PUPIL IMAGE LEARNING DEVICE, PUPIL POSITION DETECTION DEVICE, AND PROGRAM THEREFOR

PUPIL IMAGE LEARNING DEVICE, PUPIL POSITION DETECTION DEVICE, AND PROGRAM THEREFOR

机译:学生图像学习设备,学生位置检测设备及其程序

摘要

PROBLEM TO BE SOLVED: To detect a pupil position with simple constitution even when a pupil enlarged image of high resolution cannot be photographed, according to the present invention.;SOLUTION: In a pupil detection device 1, a pupil image learning device 2 comprises: an image feature quantity calculation unit 20 which calculates an image feature quantity about an edge; and a pupil image determination model generation unit 26 which generates a pupil image determination mode through machine learning on the image feature quantity, and a pupil position detection device 4 comprises: a pupil candidate setting unit 40 which sets a pupil candidate region in a face image having a face region of a person cut out; an image feature quantity calculation unit 44 which calculates an image feature quantity about an edge for a window region set in the pupil candidate region; a pupil image determination unit 46 which inputs the image feature quantity of the window region to the pupil image determination model and determines whether the window region is a pupil image; and a pupil position calculation unit 48 which calculates a pupil position included in the window region determined as a pupil image.;SELECTED DRAWING: Figure 1;COPYRIGHT: (C)2017,JPO&INPIT
机译:解决的问题:根据本发明,即使在无法拍摄高分辨率的瞳孔放大图像时,也能够以简单的结构检测瞳孔位置。解决方案:在瞳孔检测装置1中,瞳孔图像学习装置2包括:图像特征量计算单元20,计算关于边缘的图像特征量;瞳孔图像确定模型生成单元26,其通过机器学习对图像特征量生成瞳孔图像确定模式,并且瞳孔位置检测装置4包括:瞳孔候选设置单元40,其在面部图像中设置瞳孔候选区域。切掉一个人的脸部区域;图像特征量计算单元44,计算关于在瞳孔候选区域中设置的窗口区域的边缘的图像特征量;瞳孔图像确定单元46将窗口区域的图像特征量输入到瞳孔图像确定模型,并确定窗口区域是否为瞳孔图像;瞳孔位置计算单元48计算包括在确定为瞳孔图像的窗口区域中的瞳孔位置。所选择的图:图1;版权:(C)2017,JPO&INPIT

著录项

  • 公开/公告号JP2017102748A

    专利类型

  • 公开/公告日2017-06-08

    原文格式PDF

  • 申请/专利权人 NIPPON HOSO KYOKAI NHK;

    申请/专利号JP20150236043

  • 发明设计人 MOCHIZUKI TAKAHIRO;SAITO MASAHIRO;

    申请日2015-12-02

  • 分类号G06T7/00;G06T1/00;

  • 国家 JP

  • 入库时间 2022-08-21 13:58:22

相似文献

  • 专利
  • 外文文献
  • 中文文献
获取专利

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号