摘要
在疲劳驾驶研究中,人脸图像的提取至关重要,所提取图像的质量直接影响到后续图像处理和最终疲劳度判定。通过肤色采样统计和聚类分析,确定一种在YCb Cr空间下的高斯模型肤色分割方法,即把图像转换到YCb Cr颜色空间,压缩肤色分布范围,使图像亮度和色度分离。其后,应用聚类思想的最大类间方差Otsu算法,通过选取最佳动态阈值对肤色区域进一步细致分割,实现人脸区域的准确定位。
In the researches of image quality may influence fatigue driving, face image extraction the subsequent image processing is very important because the and eventual fatigue degree determination. By color sampling and clustering analysis, Gauss model skin color segmentation method under the YCbC, space is studied here. First the image is converted to YCbC~ color space for color range compression and image luminance and chrominance separation. Then with the Otsu segmentation method, more detailed segmentation for color region is obtained by choosing the optimal dynamic threshold to realize the accurate face region positioning.
出处
《长春工业大学学报》
CAS
2013年第4期382-386,共5页
Journal of Changchun University of Technology
基金
吉林省教育厅"十一五"科学技术研究项目(2010295)
关键词
人脸检测技术
颜色空间
肤色分割
高斯模型
二维Otsu算法
face detection technology
color space
skin color segmentation
Gauss model
two-dimensional Otsu algorithm.
