基于视觉的驾驶员疲劳特征提取方法-《计算机技术与发展》

文章信息/Info

Author(s):: JIN Bing-ling; ZHANG Zhen; ZHANG Zi-yao; School of Electrical and Mechanical Engineering and Automation,Shanghai University,Shanghai 200072,China

Keywords:: visual information; fatigue detection; fatigue feature extraction; AdaBoost algorithm; adaptive threshold

摘要:: 对驾驶员疲劳检测方法进行了研究,对近年来疲劳检测的研究现状进行了简要介绍,针对驾驶员疲劳检测提出了一种基于视觉的疲劳特征提取方法.对采集图像进行预处理从而提高图像对比度,利用AdaBoost算法定位人脸和眼睛区域,采用自适应阈值方法分割唇色和肤色,提取嘴唇区域;通过计算眼睑和瞳孔区域像素个数占眼部区域像素总个数的比值和嘴部区域的宽高比,分别判断眼睛和嘴巴的开闭状态,从而提取出PERCLOS特征、眨眼频率和哈欠频率等面部疲劳特征,通过疲劳特征可以进一步判断驾驶员的疲劳状态.实验结果表明,该方法可以准确定位眼部和嘴部区域,判断出视频中人员眨眼和打哈欠的次数,实现在复杂背景下驾驶员的疲劳特征提取,为后续驾驶疲劳分析提供重要的依据.

Abstract:: The method of driver fatigue detection is studied,and present status of fatigue detection is introduced chiefly. We propose a driver fatigue feature extraction method based on vision. First of all,the preprocessing of captured image is performed to improve the image contrast. After that,AdaBoost algorithm is used to locate face and eye area,and adaptive threshold method to segment lip color and skin color for extraction of lip area. By calculating the ratio of the number of pixels in the eyelid area and pupil area to the total number of pixels in the eye area and the ratio of the width and height of the mouth area,the open and closed states of the eyes and the mouth are respectively judged,so the PERCLOS feature,blink frequency and yawn frequency and other facial fatigue features are extracted to further judge fatigue state of the driver. The experiment shows that this method can accurately locate eye area and mouth area,determine the frequency of people blinking and yawning in the video,realize the extraction of fatigue features under the complex natural environment,which provides necessary support for the future analysis of driver fatigue.