一种基于眼部状态的疲劳驾驶检测方法

针对传统的疲劳驾驶检测准确率低和实时性差的问题,提出了一种基于眼部状态的疲劳驾驶检测方法.利用CCD相机实时获取驾驶员的脸部图像,采用直方图均衡化增强图像的对比度;通过改进的cascade(Hear分类器)的人脸检测算法检测出脸部区域;利用OTSU阈值分割和形态学运算提取人眼区域,根据人眼的宽高比判定眼睛的闭合程度;依据PERCLOS-P80原理和眨眼频率判断驾驶员的疲劳状态.实验结果表明:改进的人脸检测算法对每帧图像的检测时间约为45ms,在人脸检测速度上提高了2.3倍,为整个疲劳驾驶检测节省了大量的时间.研究疲劳驾驶检测方法检测一帧图像的时间约为65ms,而且在不同的光照强度下的检测均有较高的准确率,满足疲劳驾驶检测对准确性和实时性的要求.

A Fatigue Driving Detection Method Based on Eye State

According to the problem of low accuracy and poor real-time with traditional fatigue driving detection meth-od,a fatigue driving detection method based on eye state was proposed. Using the CCD camera to capture face image of driver in real time,using histogram equalization to enhance image contrast;The face area was detected by improved face detection method based on cascade;Extracting the eye area with OTSU threshold and morphology operation,The eye closure degree is determined by the aspect ratio of eye;Using PERCLOS-P80 and blink frequency to judge whether Driver is fatigue. Experimental results show that Detecting a frame needs about 45ms with Improved face de-tection algorithm, The face detection speed increased by 2.3 times, to save a lot of time for the entire fatigue driving detection. The fatigue driving detection method in this paper needs about 65ms per frame, and it has high accuracy at different light intensities,which meets the requirements of the precision and real-time for the fatigue driving detection.