Abstract: The detection of driver’s fatigue has broad application prospects and realistic meanings in numerous fields.In order to overcome the influence of the different figure with different drivers，the deflection and inclines of driver’s face，Combining the space land characteristic of the internal form of human eye，the Pseudo-Zernike moments of the land varied characteristic of the internal form of human eye，the technology of template matched，and using the sorter with the lattice degree of nearness to judge the eye state.The results show that the accuracy of this method is proved high and slightly influenced by the factor of environmental，the velocity also meets the demands of the video’s information processing.

摘要： 人眼状态判断在驾驶员疲劳检测等众多领域有着广阔的应用前景与重要的现实意义。为了克服不同驾驶员眼睛形状的差异以及在行驶过程中面部偏转、倾斜的影响，结合人眼内部形状空间域特征、人眼内部形状变换域的伪Zernike矩特征和模板匹配技术，采用基于格贴近度的分类器对人眼状态进行判断。实验结果表明该方法准确度高、受外在因素影响小，计算速度可以满足视频信息处理的需要。