您的位置:首页  > 论文页面

基于表观特征的层级式驾驶人注视状态检测

发表时间:2022-03-31  浏览量:512  下载量:48
全部作者: 肖磊,张明恒,刘晨,刘朝阳
作者单位: 大连理工大学汽车工程学院;大连理工大学工业装备结构分析国家重点实验室
摘 要: 视线方向是驾驶人注意力状态的重要表征参数之一,其检测可靠性对于车辆安全驾驶状态的实时监测具有重要作用。视线方向由驾驶人面部朝向和眼动状态两部分确定,基于单一状态特征的检测方案在检测可靠性方面还有待进一步提升。基于此,在对驾驶人视线方向形成表征要素分析的基础上,提出了一种基于视觉表观特征的两阶段视线方向检测模型,以实现对驾驶人视线方向的可靠估计。其中,基于眼睛、嘴部特征所形成的“三角”几何特性变化可实现对面部朝向的快速、准确定位;基于该定位,根据眼球对环境光照的反射特性,结合瞳孔和反射点位置所构建的眼动矢量实现驾驶人视线方向的精确检测。试验结果表明,本文所提出的两阶段驾驶人视线检测模型相对单一状态特征检测方法在检测准确性方面具有较大优势,其检测效率具有良好的车载应用潜力。
关 键 词: 公路运输;视线方向;注意力状态;主动安全;舱内感知
Title: Hierarchical driver’s fixation state detection based on apparent features
Author: XIAO Lei, ZHANG Mingheng, LIU Chen, LIU Zhaoyang
Organization: School of Automotive Engineering, Dalian University of Technology; State Key Laboratory of Structural Analysis for Industrial Equipment, Dalian University of Technology
Abstract: The gazing direction is one of the critical characteristic parameters in the driver’s attention states. Its detection reliability plays a vital role in the real-time monitoring of safe driving. Since the gazing direction is determined by the driver’s face orientation and the eye movement state, the detection scheme based on single state feature needs to be further improved on detection reliability. Therefore, based on the analysis of the representational elements of driver’s vision direction formation, a two-stage vision direction detection model based on visual apparent features is proposed to detect driver’s fixation state detection reliably. Among them, the “triangle” geometry based on eye and mouth features can locate facial orientation quickly and accurately; based on this, the precise detection of the driver’s gazing direction is further realized according to the reflection characteristics of the eyeball to the ambient light and the eye movement vector constructed by the position of the pupil and the reflection point. The test results show that the proposed two-stage driver’s gazing detection model has a tremendous advantage in detection accuracy over the detection method based on single feature, and its detection efficiency has a good potential for vehicle application.
Key words: road transport; gazing direction; attention state; activate safety; in-cabin sensing
发表期数: 2022年3月第1期
引用格式: 肖磊,张明恒,刘晨,等. 基于表观特征的层级式驾驶人注视状态检测[J]. 中国科技论文在线精品论文,2022,15(1):82-91.
 
6 评论数 0
暂无评论
友情链接