摘要
随着高速公路的飞速扩展和汽车持有量的迅速增加，公路交通事故数量明显上升，而驾驶员疲劳驾驶是引发这些交通事故的重要因素之一。据我国交通部报道，由于驾驶疲劳造成的交通事故无论是绝对数字还是所占比例都是最高的。由于交通事故并不是在一产生驾驶疲劳时就发生，所以实时的疲劳检测系统是必不可少的。因此，如何及时且有效的检测出司机的驾驶疲劳从而减少事故的发生，就成为一个非常有意义的问题。
目前国际上有不少研究疲劳驾驶检测的机构和正在开发的瞌睡检测系统，常用的疲劳驾驶检测技术有：1、通过头部位置传感器检测疲劳，由于头部运动的特征和疲劳相关性不强，因此准确率不高；2、瞳孔检测法，通过检测瞳孔直径的变化趋势估计疲劳的程度，该测量方法是一种非实时的、需佩带特殊装置的测量方法，只能用来估计疲劳的趋势；3、EEG测量法，该方法由于需要佩戴电极给被测者带来很大的精神、生理负荷，因此缺乏实用性；4、观测眼睑闭合以检测疲劳，眼睑闭合已经被证明为检测疲劳驾驶的最有效的和最有意义的特征之一。
本文分析对比了当前国内外较为流行的各种检测方法并研究了其技术关键和难点后，提出了基于计算机视觉的、无接触、无生理负荷的驾驶员疲劳检测方法，建立了基于DSP的实时驾驶员疲劳检测系统，实时采集、处理并显示图像，提取驾驶员的眼睛状态特征，以眼睛状态特征为依据，判断驾驶员是否有疲劳发生。
本论文的主要研究内容：
1、深入探讨了各种检测方法的意义、现状及原理，从分析这些检测方法中我们认为对驾驶员人体特征信号检测比较客观、准确、易行，提出了一种非接触性、全天候、实时的疲劳检测系统。
2、对疲劳检测系统设计硬件部分进行了选择，采用TI公司的TMS320DM642作为DSP处理芯片、TVP5150和SAA7121作为视频口解码和编码芯片、TL16C752B作为异步串口收发器以及一系列外围器件搭建了整个硬件系统。
3、对系统初始化、DSP／BIOS的启动以及视频口、串口等底层驱动进行了设计，使得所用的各部分硬件寄存器、整个系统初始化到一个适当的环境，为加载程序、实现算法等后续工作做好准备。
4、对具体算法进行了详细分析，分别采用肤色的空间聚类特性和背景更新的方法来定位人脸，提出了在红外光照射下利用跟踪反射点的方法快速定位跟踪人眼，利用卡尔曼滤波估计反射点的位置和闭眼状态，并结合不动点处理方法来解决闭眼的问题，采用了阈值分割的方法分离出瞳孔的图像，使用了累计直方图法二值化眼睛区域，提出了空间网格法把空间分割成网状，然后调用云台预置位的方法控制云台的运动，最后采用疲劳的各种生理特性与疲劳相关性来判定疲劳程度，证明了方法的有效性。
With the high development of the highway and the quick increase of the number of people having car, the number of traffic accidents are increasing obviously. It is reported by Ministry of Communications that both absolute number and proportion of our country's traffic accidents are highest in the world. Since accident doesn't happen as soon as fatigue takes place, real time fatigue detection system is absolutely necessary. Therefore, how to supervise and avoid fatigue driving efficiently is one of the significant problems.
Some organizations do the reseach of detecting drowsy driver, some methods of detecting drowsy driver by far are: 1. Using head position sensor to detect fatigue.While, the character of head moving is not well related to fatigue, so the accuracy rate is not high. 2. Measure pupil size: estimating the degree of fatigue by measuring the change of pupil diameter.It is not a real-time method, and the driver has to wear a special equipment.So, it is only used to estimate the trend of sleepiness. 3. Using EEG to measure the level of fatigue: it is the earliest method of detecting the fatigue, but it requires installing electrodes to the subject, which take great pale of psychological and physiological burden to them.It is lack of practice. 4 Using the eye closure to detect drowsiness, which has been proved to be the most effective character.
In this thesis, we investigate and contrast the principles of current methods and technologies for driver fatigue detection, and analyze the key problems and difficulties of these techniques, propose a quick and effective driver fatigue method based on computer vision, and establish a real-time driver fatigue detection platform of DSP. It can capture, process and display the image in real time. Further more, the system can
obtain the eye feature which is used to estimate the fatigue.
The main contents of this thesis are as follows:
1 Analyse the key technique and difficulties of detecting system home and abroad,
discuss the principle deeply, based on our analysing the detecting methods we find that the driving body feature detecting is objective, exact and practical, propose a novel and practical fatigue detecting system.
2 Describe the hardware of fatigue detecting system, use TMS320DM642 of TI company as processing chip, TVP5150 and SAA7121 as video decoder and encoder, TL16C752B as dual universal asynchronism receiver/transmitter and other periphery chips composing the whole hardware system.
3 Introduce the system intialization, DSP/BIOS startup and bottom driver such as video driver and UART driver, which can intialize the hardware registers and whole system to be ready for lasting work such as loading program, realizing algorithm and so on.
4 Analyse the algorithm detailedly, using skin distribution and background update to locate face, tracking reflection point method under infrared to track and locate eyes, Kalman filter to estimate the location of reflection point and eye-closed state, stable point mothed to solve eye-closed state, threshold division to obtain the pupil image, accumulated histogram to binarize eye area, space gridding method to control pan-and-tile movement, the relativity between fatigue physiological character and fatigue to judge fatigue degree to prove the method validity.
引文
1. Taub JM. Effects of habitual variations in napping on psychomotor performance, memory, and subjective states. International Journal of Neuroscience,1979, 9: 97-112.
2. Taub JM., Tanguay PE., Rosa RR. Effects of aftemoon naps on physiological variables, performance and self-reported activation. Biological Psychology, 1977,5: 191-210.
3. Sherry P. Job stress and fatigue in railroad dispatchers. Unpublished report. 1998
4. Taub JM, Tanguay PE, Clarkson D. Effects of daytime naps on performance and mood in a college student population. Journal of Abnormal Psychology, 1985: 210-217
5．钱录庆．日本发明防止驾驶员打瞌睡系统．南方农机，2004，5．
6．Volvo发明疲劳驾驶警示器[EB／OL]．http://www.nettx.com.cn/AutoNews/news/2006-02-20/0736168296.shtml,2006
7. Shi J, Wu YP. Identification and Analysis of Driver's Active Safety Factors Journal of Shanghai Jiaotong University. 2000,34(4)
8．王荣本，余天洪，郭烈，等．基于机器视觉的车道偏离警告系统研究综述[J]．汽车工程，2005，27(4)：463-466．
9．顾柏园，王荣本，余天洪，等．基于视觉的前方车辆探测技术研究方法综述[J]．公路交通科技，2005，(10)：114-119．
10．郭克友，储江伟，王荣本．驾驶员眼部状态识别方法的研究[J]．系统工程与电子技术，2003，(10)：1186-1188
11．杜志刚，蒋宏，潘晓东．眼动仪在道路交通安全与环境评价中的应用．第三届全国公路科技创新高层论坛论文集(下册)2006．
12. Louis Tijerina, Mark Gleckler, Duane Stoltzfus, Scott Johnston, TRC inc.;Michael J. Goodman, NHTSA; Walter W. Wierwille, Center for Transportation Research, VA Polytechnic Institute and State University(March 1999). Technical Report: A Preliminary Assessment of Algorithms for Drowsy and Inattentive Driver Detection on the Road. National Technical Information Service, Springfield, Virginia 22161.
13. Kecklund G, Akerstedt T. Sleepiness in long distance truck driving: an ambulatory EEG study of night driving. Stockholm, Sweden: Department of Stress Research and National Institute for Psychosocial Factors and Karolinska Institute. 1998; 36(9): 1007.
14. Stampi C, Aguirre A., Macchi M. . Relationship between pupilary/oculomotor measures and physiological (EEG) and behavioral indices of sleepiness. Sleep Research, 1995, 24a, 454.
15. Louis T, Mark G, Duane S . A Preliminary Assessment of Algorithms for Drowsy and Inattentive Driver Detection on the Road. DOT HS808(TBD) March 1999. 16. Philip W. K, Roger D.J Development of Driver Alertness Detection System Using Overhead Capacitive Sensor Array. http://internet.cybermesa.com/~rfrye/complexica/drowsy.htm.
17. Ji Q, Zhu ZW, Lan, P. Real-time nonintrusive monitoring and prediction of driver fatigue. IEEE Transactions on Vehicular Technology. 2004; 53: 1052-1068.
18. ITS-ASV, Mistsubishi ASV2,http://www.mitsubishi-motors.co.jo/docs4/asv2/its /index.html
19. AWAKE-System for Effective Assessment of Driver Vigilance and Warning According to Traffic Risk Estimation,http://www.awake-eu.org/index.html
20. Dingus TA, Hardee LH, Wielwille WW, Detection of Drowsy and Intoxicated Drivers based on Highway Driving Performance Measures. 1985, IEOR Department Report #8402. Vehicle Simulation Laboratory, Human Factors Groups, Virginia Polytechnic Institute and State University, Blacksburg Virginia.
21. Peters RD, Kloeppel E, Alicandri E. Effects of Partial and Total Sleep derivation on Driving Performance. US Department of Transportation, Federal Highway Administration, 1999, Publication No. FHWA-RD-94-046.
22. Texas Instrument Incorporated.TMS320DM642 Evaluation Module Technical Reference [M]. 2003.
23．李方慧,王飞,何佩琨．TMS320C6000系列DSPs原理与应用[M]．北京：电 子工业出版社，2002．
24．赵训威，基于MS320C6200系列DSP芯片的应用与开发[M]．北京：人民邮电出版社，2002．
25．徐志军，徐光辉．CPLD／FPGA的开发与应用[M]．北京：电子工业出版社，2002．
26. Texas Instrument Incorporated. TMS320C64x DSP Two-Level Internal Memory Reference Guide[M].2002.
27. Texas Instrument Incorporated.TMS320C64x EDMA Performance Data [M].2004.
28. Texas Instrument Incorporated.TMS320C64x DSP Video Port/VCXO Interpolated Control (VIC) Port Referenc Guide.2003
29. Texas Instument Incorporated. TVP51510A/TVP5150AM1 Ultralow Power NTSC/PAL/SECAM Video Decoder With Robust Sync Detector.2004
30. Phililps Semiconductors. SAA7120H; SAA7121H Digital video encoder.2002.
31. Texas Instrument Incorporated.TL16C752B 3.3-V DUAL UART WITH 64-BYTE FIFO.2000.
32. Texas Instrument Incorporated. DSP/BIOS User's Guide, SPRU423, 2002.
33. Texas Instrument Incorporated. The TMS320DM642 Video Port Mini-Driver. SPRA918A.August 2003.
34. Texas Instrument Incorporated. TMS320C6000 DSP Inter-integrated Circuit(I2C) Module Reference Guide.SPRU175.October 2002.
35. Texas Instruments Incorporated.TMS320C6000 Programmer's Guide,April 2001.
36. Texas Instruments Incorporated TMS320C6000 Optimizing C/C++ Compiler User's Guide, April 2001.
37. Texas Instruments Incorporated TMS320C6000 Assembly Language Tool User's Guide, April 2001.
38. Ellen M.A, Grace.R, Steinfeld.A. A User-Centered Drowsy- Driver Detection and Waming System, Proceedings of ACM Designing User Experiences 2003; 109-113
39. Grace R., Byme V. E. Legrand J. M., Gricourt D. J, et al. A Machine Vision Based Drowsy Driver Detection System for Heavy Vehicles. Proceedings of The Ocular Measures of Driver Alertness Conference. 1999; 75-86.
40. Tiesheng Wang, Pengfei Shi, Yawning Detection For Determine Driver Drowsiness, IEEE Int. Workshop VLSI Design & Video Tech. Suzhou, China, May 28～30, 2005.
41. Homg WB, Chen CY, Chang Y, Fan CH. "Driver fatigue detection based on eye tracking and dynamic, template matching", IEEE International Conference on Networking, Sensing and Control. 2004; 1:7-12.
42．姚鸿勋，刘明宝，高文，范旭彤，张洪明，吕雅娟．基于彩色图像的色系坐标变换的面部定位与跟踪法．计算机学报．2000年2月，第23卷，第2期．
43．田欣．基于不同颜色空间的肤色模型．西安科技学院学报，2001，21(4)：369-371
44. REINDERS MJ T, KOCH R W C, GERBRANDS J J. Locating facial features in image sequences using neural networks // Proceedings of the 2nd International Conference on Automatic Face and Gesture Recognition [C]. Vermont, 1996:230-235.
45．毛敏峰，张星明，郭宇聪，一种基于规则的眼睛快速定位算法，计算机工程，2004，30(9)154-156
46．金名蜚，一种新的人眼定位算法，淮阴师范学院学报(自然科学板)，2003，2(16)：237-240
47．田原，梁德群，图像中人脸和眼睛的定位及分割方法的研究，西安交通大学学报，1998，32(3)
48．韩海．从不同角度图像中检测眼睛．江汉大学学报，2002，9(3)：82-85
49. A.Pérez, M.L.Còrdoba, A.García, R.Méndez, M.L.Munoz, J.L.Pedraza, F.Sánchez. A Precise Eye-Gaze Detection and Tracking System. WSCG'2003, February 3-7, 2003, Plzen, Czech Republic.WSCG POSTERS proceedings.
50. Barton.T. Smart Driver.http://www.etrucker.corn/apps/news/article.asp.?id=39788