基于眼部特征的疲劳驾驶实时检测算法研究

【作者】 胡世锋；

【导师】 方祖华；

【作者基本信息】 上海师范大学 ， 通信与信息系统， 2010， 硕士

【摘要】 如今,汽车在世界范围内已经成为一种最普遍的交通工具。据德国一家汽车市场调研机构预测,全球汽车(包括个人用车和商用车)保有量2010年就将突破10亿辆,这种形势直接给世界的交通安全事业带来了极大的压力和挑战。疲劳驾驶已经成为了全球交通事故发生的主要诱因之一。在当前的疲劳驾驶检测技术领域,随着科技的不断进步,计算机视觉、图像处理和模式识别技术得到了进一步的发展和完善。基于驾驶员脸部特征的非接触式疲劳检测算法的研究和疲劳预警系统的开发已经成为了主流之一。本文针对这一背景,在自然光下利用视频图像处理技术研究了驾驶员人脸和眼睛状态的检测算法,参考PERCLOS疲劳检测准则研究出了一种比较准确的疲劳检测方法,从而实现了对驾驶疲劳的检测和预警。本文所做的主要工作包括:(1)针对驾驶员人脸的检测,利用YCbCr和HSV双色彩空间的肤色信息和Adaboost分类器的融合研究出了一种具有鲁棒性好、计算速度快的驾驶员人脸检测与定位算法。(2)在人脸检测与定位的基础上,基于由粗到精多级筛选的思想,利用人眼区域模板匹配算法和Hough变换瞳孔圆检测的算法实现了对人眼的定位和眼睛开闭状态认定,为后续的疲劳判断打下了理论基础。(3)在判定了人眼状态的条件下,通过对驾驶疲劳的认真分析,参照PERCLOS准则设计出了一种基于阈值分割思想的疲劳检测方法。(4)利用C语言对整个算法进行了源码实现,借助OpenCV开发平台完成了对整个检测算法的效果验证。经过实验室模拟试验表明,本文研究的这种基于眼部特征的疲劳驾驶实时检测算法具有比较好的正确率和鲁棒性,为以后更深入的研究和算法的硬件实现打下了基础。更多还原

【Abstract】 Today, the car in the world has become one of the most common means of transport. According to a German automotive market researching institution predicted that Global vehicles (including personal cars and commercial vehicles) holdings in 2010 will surpass 1 billion. This situation has brought great pressures and challenges to the world’s traffic safety directly. Fatigue driving has become one of the main incentives of the world’s traffic accidents.In the current driver fatigue detection field, as technology advances, computer vision, image processing and pattern recognition technology has been further developed and improved. The research based on driver facial features, non-contact fatigue detection algorithm and development of fatigue warning systems have become one of the mainstream.In view of this background, this paper developed a detection algorithm based on driver’s face and eye state, with the video image processing technology in natural light. Referring to PERCLOS criteria of fatigue testing, it developed a accurate fatigue detection method, and achieved the driver fatigue detection and early warning.This paper included the following contents:(1)For the driver’s face detection, it developed a driver’s face detection and localization algorithm which has good robustness and fast calculation. The algorithm syncretized the skin color information based on YCbCr & HSV color space and Adaboost classifier.(2)In the basis of human face detection and location, it achieved pupil of human eyes orientation and the state of eyes identified, using the area of human eye template matching algorithm and the Hough transform circle detection algorithm. The method based on the from coarse to fine multi-level filtering idea. It had laid a theoretical basis for follow-up to determine the fatigue.(3)In the determination of the conditions of the state of the human eyes, through careful analysis of fatigue on driving, referring to PERCLOS criteria it designed a fatigue detection method based on the idea of threshold-segmentation.(4)It implemented the entire algorithm with C language source code. Using OpenCV development platform, it completed the effect of the whole detection algorithm validation.Through laboratory simulations, it showed that based on feature of the eyes driving fatigue real-time detection algorithm has better accuracy and robustness. It laid the foundation for more in-depth researching of algorithm and hardware implementation.更多还原