驾驶员视线估计方法综述  

A Summary of Gase Estimation Methods

作　　者：张春雨[1] 郭克友[2] 蔡蕾[1] 王琪[1] 

机构地区：[1]交通部公路科学研究院智能交通关键技术交通行业重点实验室,北京100088 [2]北京工商大学机械自动化学院,北京100037

出　　处：《公路交通科技》2009年第3期139-143,共5页Journal of Highway and Transportation Research and Development

基　　金：国家高技术研究发展计划(八六三计划)资助项目(2006AA11Z114)

摘　　要：研究基于计算机视觉的视线估计方法,对驾驶员视线估计的暨有研究成果加以回顾与评述,并对可能的发展趋势进行分析。主要针对3类基于计算机视觉的方法展开论述:基于PCCR(Pupil Center Corneal Reflection)技术的视线估计方法;基于AAM模型(Active Appearance Model)的视线估计方法;基于统计模式识别的视线估计方法。总体上讲各种技术都是基于图像传感器的,很难突破图像传感器特有的鲁棒性不强、难以适应全天候工作要求的缺点,但各种技术有着各自的特点,因此也很难用统一的标准去衡量各种监测技术的优劣。本文给出了几种有代表性系统的性能比较。The latest researches of gaze estimation methods based on computer vision was reviewed and the potential development in this area was analyzed. Three kinds of gaze estimation methods based on computer vision were discussed： methods based on Pupil Center Corneal Reflection （PCCR）, methods based on Active Appearance Model （AAM） and methods based on statistical pattern recognition. Although all the methods mentioned above are based on image sensor which is difficult to overcome the shortcomings of lacking of robustness and unfit for the all-weather working conditions, these methods have their own characteristics, so it is difficult to weigh up the superiority and inferiority of these methods by a unified criterion. Comparison of the performance of current gaze estimation systems was given.

关 键 词：智能运输系统 智能监控 视线估计 安全辅助驾驶 疲劳监测 

分 类 号：U491[交通运输工程—交通运输规划与管理]