非约束条件下驾驶人眼睛检测与跟踪方法研究

详细信息    查看全文 | 推荐本文 |

英文篇名：Research on Driver Eye Detection and Tracking under Non-Restraint Conditions 作者：程文冬 ; 魏庆媛 英文作者：CHENG Wendong;WEI Qingyuan;School of Mechatronic Engineering,Xi'an Technological University;School of Mechanical Engineering,Harbin Institute of Petroleum; 关键词：眼睛检测 ; 模板匹配 ; 非约束条件 ; Kalman滤波 英文关键词：eye detection;;template matching;;non-restraint conditions;;Kalman filtering 中文刊名：XAGY 英文刊名：Journal of Xi’an Technological University 机构：西安工业大学机电工程学院;哈尔滨石油学院机械工程学院; 出版日期：2019-04-19 出版单位：西安工业大学学报 年：2019 期：v.39;No.210 基金：国家自然科学基金(51775053) 语种：中文; 页：XAGY201902016 页数：8 CN：02 ISSN：61-1458/N 分类号：87-94

摘要

为了抑制驾驶人头部姿态与眼球运动等非约束条件对眼睛检测的干扰,文中提出了一种基于合成模板匹配、融合Cam Shift与Kalman滤波的人眼检测与跟踪算法。将驾驶人左、右眼睛样本人工合成模板对眼睛区域进行匹配检测,采用基于差值平方和算法的相似度度量方法验证眼睛身份。通过眼睛颜色直方图反投影生成灰度概率分布图像,基于Cam Shift算法计算眼睛跟踪窗口的位置和大小,采用Kalman滤波预测眼睛区域位置,有效解决了驾驶人头部转动、眨眼等动作的干扰问题。实车试验结果表明:该算法对驾驶人人眼的平均识别率达到94.3%,平均耗时为22.8 ms,能够较好的兼顾驾驶人人眼跟踪的准确性、鲁棒性与实时性。
        In order to crush the interference with eye detection caused by non-restraint conditions,such as driver's head posture and eye movement,the paper presents an eye detection and tracking method,which is based on synthetic template matching and combined with Cam Shift and Kalman filtering.Firstly,a synthetic matching template is built up based on left and right eye samples for eye detection.Eye identity is then verified by the similarity measure method of difference sum of squares.Secondly,the raw image is converted into gray probability distribution image through color histogram back projection.The Cam Shift algorithm is then applied to estimate the position and size of the eye tracking window.To deal with the interference of head posture and eye movement,the Kalman filtering algorithm is introduced into the eye tracking system for estimating eye region center in next frame.Real vehicle experiment demonstrates that the average eye detection rate reaches 94.3% and the average calculating time is 22.8 ms.It is concluded that the proposed method has good accuracy,robustness and real-time performance.

引文

[1] 王长元,李京京,贾宏博,等.头眼运动研究方法及进展[J].西安工业大学学报,2012,32(3):173. WANG Changyuan,LI Jingjing,JIA Hongbo,et al.Overview of the Method and Progress of Eye-Head Movement Research[J].Journal of Xi’an Technological University,2012,32(3):173.(in Chinese)
    [2] 董文会,曲培树,吴晓娟,等.复杂背景下多姿态人脸图像中的人眼检测[J].山东大学学报(工学版),2006,36(3):18. DONG Wenhui,QU Peishu,WU Xiaojuan,et al.Eye Detection for Multi-pose Face Image in Complex Background[J].Journal of Shandong University (Engineering Science),2006,36(3):18.(in Chinese)
    [3] 张伟,成波,张波.驾驶人眼睛定位与跟踪算法的研究[J].汽车工程,2012,34(10):889. ZHANG Wei,CHENG Bo,ZHANG Bo.A Research on the Algorithm for Driver’s Eye Positioning and Tracking [J].Automotive Engineering,2012,34(10):889.(in Chinese)
    [4] 史慧荣,张学帅,梁彦,等.一种改进的模板匹配眼睛定位方法[J].计算机工程与应用,2004(33):44. SHI Huirong,ZHANG Xueshuai,LIANG Yan,et al.An Improved Template Matching Eye Location Method[J].Computer Engineering and Applications,2004(33):44.(in Chinese)
    [5] 程文冬,付锐,袁伟,等.驾驶人注意力分散的图像检测与分级预警[J].计算机辅助设计与图形学学报,2016,28(8):1287. CHENG Wendong,FU Rui,YUAN Wei,et al.Driver Attention Distraction Detection and Hierarchical Prewarning Based on Machine Vision[J].Journal of Computer-Aided Design & Computer Graphics,2016,28(8):1287.(in Chinese)
    [6] 万中田,冼钟业,胡明宇,等.基于Kalman预测器的多特征Camshift运动目标跟踪算法[J].武汉大学学报(工学版),2015,48(5):712. WAN Zhongtian,XIAN Zhongye,HU Mingyu,et al.Moving Target Tracking Based on Kalman Predictor and Multi-featured Camshift Algorithm[J].Engineering Journal of Wuhan University,2015,48(5):712.(in Chinese)
    [7] 姚红革,耿军雪.基于卡尔曼预测的视频目标实时跟踪[J].西安工业大学学报,2007,27(2):171. YAO Hongge,GENG Junxue.A Real-Time Tracking for Video Target Based on Kalman Prediction[J].Journal of Xi’an Technological University,2007,27 (2):171.(in Chinese)