基于机器视觉的驾驶人使用手持电话行为检测  

Detection handheld phone use by driver based on machine vision

作　　者：张波[1] 王文军[1] 魏民国 成波[1] 

机构地区：[1]清华大学汽车安全与节能国家重点实验室,北京100084

出　　处：《吉林大学学报（工学版）》2015年第5期1688-1695,共8页Journal of Jilin University:Engineering and Technology Edition

基　　金：交通运输部信息化科技项目(2012-364-835-110);国家自然科学基金项目(51205228)

摘　　要：为了对驾驶过程中使用手持电话行为实施监控和预警,提出了一种基于机器视觉的驾驶人使用手持电话行为检测方法。该方法使用ASM算法分割出面部肤色区域,并运用F-B Error算法在头部发生偏转时对肤色区域进行跟踪。在此基础上,提出用自适应肤色范围提取算法进行肤色检测,并确定手部图像候选区域。提出了CMOG特征描述手部图像,并综合使用CMOG特征与HOG特征在手部图像候选区域中确定驾驶人的手部图像。最后,运用三模式逻辑判别方法判别驾驶人使用手持电话的行为。实验结果表明:该方法能够有效检测出驾驶人在驾驶过程中使用手持电话的行为。A machine vision based method is proposed to detect handheld phone use behavior.In this method,first,Active Shape Model（ASM）algorithm is used to segment the skin-color region of the driver face,and the region is tracked by F-B error algorithm when the head moves.Second,an adaptive skin-color range extraction algorithm is proposed to detect the skin-color pixels and determine the candidates of hand images.Third,Co-occurrence Matrix of Oriented Gradients（CMOG）is proposed,which combined with Histogram Oriented Gradient（HOG）feature is used to detect the real hand image from the candidates.Finally,three modes logic discriminating method is used to identify the driver state.Experiment results show that the proposed machine vision based method can detect the handheld phone use behavior effectively.

关 键 词：信息处理技术 驾驶行为 机器视觉 手持电话 

分 类 号：TN971[电子电信—信号与信息处理]