基于几何特征的疲劳检测算法研究与实现.pdfVIP

Computer Science and Application 计算机科学与应用, 2012, 2, 246-250 doi:10.4236/csa.2012.25043 Published Online December 2012 (/journal/csa.html) Implementation of Fatigue Detection Algorithm Based on * Geometric Features Jingdan Wang, Chuanhua Jiang, Zongbo Hao School of Computer Science Engineering, University of Electronic Science and Technology of China, Chengdu Email: wjd719@ Received: Oct. 21st th th , 2012; revised: Nov. 16 , 2012; accepted: Nov. 25 , 2012 Abstract: The Adaboost algorithm based on the face of the classifier can accurately detect and positioning face, then through trying three methods to locate the human eye, such as Frame Difference and geometric feature method, extract- ing human eye image to obtain the binary image of eye by gray processing, gauss smooth, adaptive threshold. Finally, the algorithm will further accurately judge the driver is in a state of fatigue through the analysis of eyes’ feature. The experimental results show that the method of Geometric Feature has anti-jamming effect in positioning the human eye, and the eyes feature significantly can ensure Robustness and real-time. Keywords: Image Pre-Processing; Classifier; Geometric Features; Fatigue State 基于几何特征的疲劳检测算法研究与实现* 王景丹，江春华，郝宗波 电子科技大学计算机科学与工程学院，成都 Email: wjd719@ 收稿日期：2012 年 10 月 21 日；修回日期：2012 年 11 月 16 日；录用日期：2012 年 11 月 25 日 摘 要：采用基于人脸分类器的 Adaboost 算法可以准确检测并定位人脸，然后通过尝试帧差法及几何特征法来 定位人眼，提取人眼图像，进行灰度化、高斯平滑、自适应阈值等预处理，最终获取人眼二值化图像，进一步 精确到眼睛部位，分析眼睛的状态特征，判定驾驶员是否处于疲劳状态。实验结果表明几何特征定位人眼的方 法有抗干扰的效果，眼睛特征表现明显，鲁棒性好，同时也能够保证实时性。 关键词：图像预处理；分类器；几何特征；疲劳状态