基于人脸识别驾驶员疲劳检测系统设计与开发

发布时间：2018-02-03 10:53

本文关键词： 机器视觉颜色空间人脸检测 PCA 疲劳驾驶识别　出处：《大连理工大学》2016年硕士论文　论文类型：学位论文

【摘要】：在交通飞速跃进的现代社会,开车的人数增长迅速,但醉酒驾车和过累开车使得道路事故急剧增加,给人民群众的生活和社会的安全带来了巨大的隐患。所以对开车人疲劳程度的检验和辅助开车装置的研发显得越来越急迫和重要。凭借关于人脸的研究结果,本文主要是通过PCA理论和其他相关算法实现开车人脸检验,并根据检验结果辅助驾驶员安全行驶。论文的核心工作是通过算法,实现了定位开车人脸位置、通过开车人肤色特征结合坐标投影算法快速计算面部表情变化过程中表现出来的数字特征,并在后文根据这些特征判断人脸面部动作。接下来关注张开、半闭合、闭合这三个情况嘴和眼的变化进行拍照,采集到有用的高度联系的数据,通过这些数据判断驾驶员是否出现呵欠及眼睛闭眼等动作,进而帮助建立判断模型和进行系统的设计。凭借检查到的开车人的局部图像完成开车人过累状态的判别,可以帮助开车人驾车,避免事故的发生。开车人脸过累识别不但是当前研究的热点,还是机器视觉和模式识别的拓展,在考察脸局部特征的基础上,凭借分块矩阵划分的手段,结合开车人过累时所特有的境况,设法把人脸的特征器官勾勒出来,强调了关键数据,弱化了不重要数据造成的干扰,然后综合判断开车人的过累情况,这个做法效果好,准确度高,由于驾驶员疲劳识别对于速度和精度要求都很高,本系统的精度基本符合要求。本文研究的内容主要是基于肤色模型的开车人人脸检测,基于分块PCA的人脸局部特征驾驶员特征识别以及整个驾驶员疲劳识别系统的构建。
[Abstract]:In a modern society with rapid leaps and bounds in traffic, the number of people driving is growing rapidly, but drunken driving and overworked driving have led to a sharp increase in road accidents. It brings great hidden danger to the people's life and the safety of the society. So it is more and more urgent and important to test the fatigue degree of the driver and to develop the auxiliary driving device. This paper mainly through PCA theory and other related algorithms to achieve face testing, and according to the test results to assist drivers to drive safely. The core work of this paper is to achieve the location of the driving face through the algorithm. The color features of drivers and coordinate projection algorithm are used to quickly calculate the digital features in the process of facial expression change, and then to judge the facial movements according to these features. Next, focus on the opening. The changes of mouth and eyes in the three cases of semi-closure and closure were photographed, and useful highly related data were collected to determine whether the driver was yawning and closing his eyes. Then it can help to establish the judgment model and design the system. With the local image of the driver checked to complete the identification of the overtired state of the driver can help the driver drive. Face recognition is not only the focus of current research, but also the expansion of machine vision and pattern recognition. Based on the investigation of local features of face, the method of partition matrix is adopted. Combined with the unique situation of the driver when overtired, try to outline the characteristic organs of the face, emphasize the key data, weaken the interference caused by the unimportant data, and then comprehensively judge the overtired situation of the driver. This method has good effect and high accuracy. Because of driver fatigue identification, the speed and precision are very high. The precision of the system is basically in line with the requirements. The main content of this paper is based on the skin color model driver face detection. The driver feature recognition based on block PCA and the construction of driver fatigue recognition system.
【学位授予单位】：大连理工大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：U463.6;TP391.41

【相似文献】