您的位置:山东大学 -> 科技期刊社 -> 《山东大学学报(工学版)》

山东大学学报 (工学版) ›› 2018, Vol. 48 ›› Issue (5): 69-76.doi: 10.6040/j.issn.1672-3961.0.2017.424

• 机器学习与数据挖掘 • 上一篇    下一篇

基于单目摄像头的主动式驾驶行为分析算法

吴晨谋1(),方志军1,*(),黄正能2   

  1. 1. 上海工程技术大学电子电气工程学院, 上海 201620
    2. 华盛顿大学电气工程系, 华盛顿 西雅图 352500
  • 收稿日期:2017-08-29 出版日期:2018-10-01 发布日期:2017-08-29
  • 通讯作者: 方志军 E-mail:747519579@qq.com;zjfang@foxmail.com
  • 作者简介:吴晨谋(1993—),男,江苏无锡人,硕士研究生,主要研究方向为机器视觉,姿态估计与行为识别.E-mail:747519579@qq.com
  • 基金资助:
    国家自然科学基金资助项目(61461021);上海市科委地方院校能力建设资助项目(15590501300)

Active driving behavior analysis algorithm based on monocular camera

Chenmou WU1(),Zhijun FANG1,*(),Jenqneng HWANG2   

  1. 1. School of Electrical and Electronic Engineering, Shanghai University of Engineering Science, Shanghai 201620, China
    2. Department of Electrical Engineering, University of Washington, Seattle 352500, Washington, USA
  • Received:2017-08-29 Online:2018-10-01 Published:2017-08-29
  • Contact: Zhijun FANG E-mail:747519579@qq.com;zjfang@foxmail.com
  • Supported by:
    国家自然科学基金资助项目(61461021);上海市科委地方院校能力建设资助项目(15590501300)

摘要:

为了预防事故发生,提出一种以人体三维姿态估计对驾驶员行为进行识别监测的算法。利用单目摄像头获取运动中驾驶员的视频流,提取每帧图像的二维轮廓特征,与预先建立的三维人体模型的二维投影进行匹配,实时估计驾驶员上半身的姿态。根据获取驾驶员的8个骨骼节点的三维坐标,对驾驶员的行为识别分析。试验模拟驾驶员正常、单手、接听电话和疲劳/醉酒驾驶4种驾驶状态,通过骨骼节点的坐标变化,实现检测和识别驾驶员的姿态行为并给予提醒。在光线较好的情况下,与PRECLOSE(percent eye closure)算法相比,该算法的误检率降低了24.24%。

关键词: 交通事故, 驾驶状态, 单目摄像头, 姿态估计, 行为识别, 模拟退火

Abstract:

In order to prevent accidents, an algorithm for recognizing and monitoring the driver′s behavior based on the three-dimensional pose estimation of the human body was proposed. A monocular camera was used to capture the video stream of the driver in motion, the two-dimensional contour features of each frame of the image was extracted, and the two-dimensional projection of the pre-established three-dimensional human body model was matched to estimate the attitude of the driver′s upper body in real time. Based on the three-dimensional coordinates of the driver′s eight skeletal nodes, the driver′s behavior was identified and analyzed. Four driving states of driver′s normal, one-handed, answering calls and fatigue/drunk driving were simulated. Through the coordinate changes of the skeletal nodes, the gesture behavior of the driver could be detected and recognized, and the driver could be given reminders. When the light was enough, the algorithm could reduce the false detection rate by 24.24% compared with the PRECLOSE algorithm.

Key words: traffic accident, driving condition, monocular camera, pose estimation, behavior recognition, simulate anneal

中图分类号: 

  • TP391

图1

算法流程图"

图2

边缘特征示例"

图3

肤色特征示例"

图4

投影特征示例"

图5

模拟退火算法流程图"

图6

4种驾驶行为"

图7

长袖状态下人体上身姿态估计"

图8

短袖状态下人体上身姿态估计"

图9

4种驾驶行为姿态估计"

表1

正常驾驶状态下各节点的三维坐标"

部位 x y z
7.302 7.639 8.699
右肩 5.985 8.113 5.019
左肩 7.258 6.809 1.800
躯干 16.369 17.921 11.958
右肘 29.141 31.112 9.220
左肘 24.823 22.337 5.668
右手 34.776 28.653 14.690
左手 27.992 36.542 14.340

表2

单手驾驶状态下各节点的三维坐标"

部位 x y z
4.998 5.229 5.954
右肩 4.097 5.553 3.436
左肩 4.968 4.661 1.232
躯干 11.205 12.267 8.185
右肘 19.947 21.296 6.311
左肘 26.669 33.323 3.961
右手 19.161 25.013 9.816
左手 23.804 19.613 10.055

表3

接电话下各节点的三维坐标"

部位 x y z
4.302 6.639 6.699
右肩 4.985 5.113 3.019
左肩 4.258 4.809 1.821
躯干 11.369 12.921 8.958
右肘 20.141 31.112 6.922
左肘 16.991 15.290 3.880
右手 7.776 11.653 4.690
左手 20.112 21.39 7.816

表4

醉酒/疲劳状态下各节点的三维坐标"

部位 x y z
1.084 1.079 0.473
右肩 0.980 1.510 0.467
左肩 0.970 2.245 0.063
躯干 5.914 6.107 3.806
右肘 7.958 5.841 8.434
左肘 7.673 7.564 8.418
右手 9.266 11.352 7.505
左手 8.420 10.527 6.721

图10

双手间的距离与肩宽的比值"

图11

手与手/手与头之间的坐标距离"

图12

头部x轴坐标变化曲线"

图13

头部y轴坐标变化曲线"

图14

头部z轴坐标变化"

表5

本研究算法与PERCLOS的比较"

算法 模拟疲劳驾驶次数 检测疲劳驾驶次数 误检次数
本研究算法 12 13 1
PERCLOS 12 16 4

表6

算法在不同光线环境下的效果"

性能指标 总帧数 真实疲劳驾驶次数 检测疲劳驾驶次数 真实打电话次数 检测打电话次数 真实单手次数 检测单手次数
光线较好 15 680 3 3 3 3 3 2
光线较差 10 079 3 2 3 1 3 4
1 国家统计局.中华人民共和国2016年国民经济和社会发展统计公报[R/OL].[2017-02-28]. http://www.stats.gov.cn/tjsj/zxfb/201702/t20170228_1467424.html.
2 NOBE S A, WANG F Y. An overview of recent developments in automated lateral and longitudinal vehicle controls[C]//Proceedings of IEEE International Conference on Systems, Man, and Cybernetics. Tucson, USA: IEEE, 2002: 3447-3452.
3 TAKEI Y, FURUKAWA Y. Estimate of driver's fatigue through steering motion[C]//Proceedings of IEEE International Conference on Systems, Man and Cybernetics. Waikoloa, USA: IEEE, 2006: 1765-1770.
4 DAS D , ZHOU S , LEE J D . Differentiating alcohol-Induced driving behavior using steering wheel signals[J]. IEEE Transactions on Intelligent Transportation Systems, 2012, 13 (3): 1355- 1368.
doi: 10.1109/TITS.2012.2188891
5 SIMONS R , MARTENS M , RAMAEKERS J , et al. Effects of dexamphetamine with and without alcohol on simulated driving[J]. Psychopharmacology, 2012, 222 (3): 391- 399.
doi: 10.1007/s00213-011-2549-0
6 METS M A , KUIPERS E , LM D S D , et al. Effects of alcohol on highway driving in the STISIM driving simulator[J]. Human Psychopharmacology Clinical & Experimental, 2011, 26 (6): 434- 439.
7 SAHAYADHAS A , SUNDARAJ K , MURUGAPPAN M . Detecting driver drowsiness based on sensors: a review[J]. Sensors, 2012, 12 (12): 16937- 16953.
doi: 10.3390/s121216937
8 INGRE M , AKERSTEDT T , PETERS B , et al. Subjective sleepiness, simulated driving performance and blink duration: examining individual differences[J]. Journal of Sleep Research, 2010, 15 (1): 47- 53.
9 KE S R, ZHU L J, HWANG J N, et al. Real-time 3D human pose estimation from monocular view with applications to event detection and video gaming[C]//Proceedings of 2010 Seventh IEEE International Conference on Advanced Video and Signal Based Surveillance (AVSS). Boston, USA: IEEE, 2010: 489-496.
10 SUN Ruizhi.3D front-view human upper body pose estimation using single camera[D]. Seattle: University of Washington, 2013.
11 FELZENSZWALB P , MCALLESTER D , RAMANAN D . A discriminatively trained, multiscale, deformable part model[J]. Cvpr, 2008, (8): 1- 8.
12 HAN F. A two-stage approach to people and vehicle detection with HOG-based SVM[C]//Proceedings of Proc. of Workshop on Performance Metrics for Intelligent Systems. Piscataway, NJ: IEEE, 2006: 133-140.
13 VIOLA P, JONES M. Rapid object detection using a boosted cascade of simple features[C]//Proceedings of the 2001 IEEE Computer Society Conference on Computer Vision and Pattern Recognition.Kauai, Hawaii: IEEE, 2001: 511.
14 LIENHART R, MAYDT J. An extended set of Haar-like features for rapid object detection[C]//Proceedings of International Conference on Image Processing, Rochester. NY, USA: IEEE, 2002.
15 LI H, LIN Z, SHEN X, et al. A convolutional neural network cascade for face detection[C]//Proceedings of 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR 2015). Boston, MA, USA: IEEE, 2015: 5325-5334.
16 CHIEN Y . Pattern classification and scene analysis[J]. The Library Quarterly: Information, Community, Policy, 1974, 44 (3): 462- 463.
17 YOON I , KIM S , KIM D , et al. Adaptive defogging with color correction in the HSV color space for consumer surveillance system[J]. IEEE Transactions on Consumer Electronics, 2012, 58 (1): 606- 607.
18 SCHWARZ M W , COWAN W B , BEATTY J C . An experimental comparison of RGB, YIQ, LAB, HSV, and opponent color models[J]. Acm Transactions on Graphics, 1987, 6 (2): 123- 158.
doi: 10.1145/31336.31338
19 VECCHI M P , KIRKPATRICK S . Global wiring by simulated annealing[J]. IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems, 2004, 2 (4): 215- 222.
20 TRUTSCHEL U, SIROIS B, SOMMER D, et al. PERCLOS: An alertness measure of the past[C]//Proceedings of the Sixth International Driving Symposium on Human Factors in Driver Assessment, Training and Vehicle Design. California, USA: IEEE, 2011: 172-179.
[1] 史青宣,王谦,田学东. 基于层叠的部件轨迹片段模型的视频人体姿态估计[J]. 山东大学学报(工学版), 2018, 48(2): 14-21.
[2] 左俊彦, 张建国, 钟涛. 基于Canny检测的股骨边缘轮廓连接算法[J]. 山东大学学报(工学版), 2015, 45(3): 65-72.
[3] 张潇丹,赵力,邹采荣*. 一种改进的混合蛙跳算法求解有约束优化问题[J]. 山东大学学报(工学版), 2013, 43(1): 1-8.
[4] 王新沛1,刘常春1*,白曈2. 基于均值距离的图像分割方法[J]. 山东大学学报(工学版), 2010, 40(4): 36-41.
[5] 王振树 李林川 李波. 基于粒子群与模拟退火相结合的无功优化算法[J]. 山东大学学报(工学版), 2008, 38(6): 15-20.
Viewed
Full text


Abstract

Cited

  Shared   
  Discussed   
[1] 孟健, 李贻斌, 李彬. 四足机器人跳跃步态控制方法[J]. 山东大学学报(工学版), 2015, 45(3): 28 -34 .
[2] 何东之, 张吉沣, 赵鹏飞. 不确定性传播算法的MapReduce并行化实现[J]. 山东大学学报(工学版), 0, (): 22 -28 .
[3] 张建明, 刘泉声, 唐志成, 占婷, 蒋亚龙. 考虑剪切变形历史影响的节理峰值剪切强度准则[J]. 山东大学学报(工学版), 0, (): 77 -81 .
[4] 王换,周忠眉. 一种基于聚类的过抽样算法[J]. 山东大学学报(工学版), 2018, 48(3): 134 -139 .
[5] 肖乔, 裴继红, 王荔霞, 龚志成. 基于多通道Gabor滤波模糊融合的遥感图像舰船检测[J]. 山东大学学报(工学版), 0, (): 29 -35 .
[6] 马相明, 孙霞, 张强. 轮式装载机典型作业工况构建与分析[J]. 山东大学学报(工学版), 0, (): 82 -87 .
[7] 王兰忠, 孟文杰. 远程教育中视频感知加密算法[J]. 山东大学学报(工学版), 2015, 45(4): 40 -44 .
[8] 梁泽华,崔耀东,张雨. 有顺序依赖损耗的一维下料问题[J]. 山东大学学报(工学版), 2018, 48(3): 75 -80 .
[9] 朱登元, 姚占勇, 葛守仁. 柱式桥台前坡改造数值分析与现场检测[J]. 山东大学学报(工学版), 2015, 45(3): 86 -94 .
[10] 叶明全,高凌云,万春圆. 基于人工蜂群和SVM的基因表达数据分类[J]. 山东大学学报(工学版), 2018, 48(3): 10 -16 .