视频监测中的驾驶人面部特征点跟踪

Tracking driver's facial feature point in video monitoring

由于具备多尺度多方向特性,Gabor小波常常被用于纹理分析及特征提取等方面的研究,并在驾驶人疲劳状态监测技术领域中获得较多的应用。首先介绍了Gabor小波的基本原理及使用方法,然后结合实例重点讨论驾驶人面部特征点跟踪的技术环节,包括Gabor小波的参数确定方法和待跟踪点特征向量提取方法,最后利用最大相识原则确定了3个备选结果。针对不同的图像条件,提出了Gabor小波在实际应用时需要注意的合理建议。实践证明,跟踪算法运算结果准确率较高,效果非常理想,能够为驾驶人疲劳状态分析打下良好的基础。

With multi-scale and multi-directional characteristics,Gabor wavelet is often used for texture analysis and feature extraction studies,and obtained more applications in the field of driver fatigue monitoring technology.First,the basic principles of Gabor wavelet and using methods are introduced,and then with examples technology segments of driver facial feature point tracking are focused on,including the method of determining Gabor wavelet parameters and the method of extracting the feature point＇s eigenvector.And finally the three candidates with the maximum similarity rules are identified.For the different image conditions,reasonable proposals in the actual application of Gabor wavelet are proposed.Practice has proved,the Gabor wavelet track-ing algorithm has higher accuracy.The effect is very ideal,and lays a good foundation for the driver fatigue analysis.