摘要
煤矿巷道监控图像中脸部分割的难度比较大.在典型监控角度方向采集了巷道中矿工的监控图像;在HSV颜色模型的各分量中分析了矿工脸部肤色的统计特征;在色度分量空间,采用上限和下限阈值的方法分割出肤色区域像素和相似色度的背景像素;对脸部区域的二值图像进行平滑处理和分割,大幅度降低背景像素的影响,缩小脸部区域的范围.实验结果表明,此方法计算简单、速度快,能成功地分割巷道监控图像中的脸部区域.
It is very difficult to segment the faces in coal mine laneway surveillance images. The miners surveilled images are sampled at typical surveilling angles in laneway. The statistic feature of minersface color is analyzed separately in tri-component of HSV color model. The pixels of skin color area and the background ones that seem to shin color ones are segmented based on the segmentation method with upper and lower threshold value. The influence of background pixels is decrease observably and the region of face is reduced, after the binary images of faces is smoothed and segmented. The results of experiments show that the method above is simple and celerity in computing and can segment the faces area in laneway surveillance images successfully.
出处
《小型微型计算机系统》
CSCD
北大核心
2009年第5期1000-1003,共4页
Journal of Chinese Computer Systems
基金
中国矿业大学科技基金项目
关键词
巷道监控图像
颜色模型
分割
平滑
laneway surveillance images
color model
segmentation
smoothing
作者简介
陈伟,男,1978年生,博士,讲师,研究方向为图像处理与模式识别、矿井监控与通信; E-mail:chenwei13046@yahoo.com.cn
孙统风,男,1977年生,博士研究生,讲师,研究方向为图像处理与识别;
杨小冬,男,1972年生,博士,讲师,研究方向为信号与信息处理.