基于颜色空间和面部特征的驾驶员面部定位方法  

Driver's Face Positioning Method Based on Color Space and Face

作　　者：符凯[1] 葛如海[1] 符鸿玉[1] 金桥[1] 

机构地区：[1]江苏大学汽车与交通工程学院车辆工程系,江苏镇江212013

出　　处：《重庆交通大学学报（自然科学版）》2013年第1期95-98,共4页Journal of Chongqing Jiaotong University(Natural Science)

摘　　要：介绍了一种快速检测和定位驾驶员面部区域的方法。首先将驾驶员面部图像映射到YCbCr颜色空间,建立肤色高斯模型;然后通过似然图自适应阈值分割获得驾驶员面部区域,对获得的二值化面部图像进行形态学操作消除毛刺和小孔;最后对处理后的面部图像进行灰度积分投影,结合面部特征获得驾驶员的面部区域并框定。并试验证明了方法的实时有效性。A fast method for the detection and location of the driver's facial area was introduced.Firstly the image was mapped into YCbCr space and the skin Gaussian model was built.Secondly driver's face area was got by likelihood figure adaptive threshold segmentation.The processed binarization image should be carried out morphology operation to eliminate burrs and holes.Finally the facial image was processed with gray integral projection to get driver's face area.The test shows this method has effectiveness and timeliness.

关 键 词：人脸检测 肤色模型 YCBCR颜色空间 形态学 灰度积分投影 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]