Abstract: Based on the video of driving fatigue detection，a new algorithm of eye localization and eye state recognition in driving fatigue detection is proposed.This algorithm uses the Gabor filter to smooth the image and pick up the terrain feature of the images.Then SVM is used to validate the eye area.Finally，SVM is used to recognize the eye state.The robustness of the algorithm is demonstrated under different eye states and facial appearances.It can determine the eye state qualitatively and quantitatively.The result of the algorithm provides necessary data for the future analysis of driving fatigue.

摘要： 基于驾驶疲劳研究中的视频图像，提出一种新的适用于驾驶疲劳检测的人眼定位及状态判别算法。定位算法利用Gabor小波变换对图像进行预处理，提取基于图像的地形图特征，并利用支持向量机的方法对候选特征区域进行验证，然后在定位区域对人眼状态进行判断。经实验验证，该方法可以定位睁眼及闭眼图像，并可以定性和定量判断人眼状态。采用该方法获得的人眼状态信息将为后续的驾驶疲劳分析提供重要的数据。