基于YCbCr模型的巷道监控中矿工脸部图像识别

Miners' Face Images Identification in Monitor Images of Mine Gateway Based on YCbCr Model

矿工脸部区域的分割对于快速检索煤矿监控图像存档资料中矿工图像有十分重要的实际意义。在典型监控角度方向采集了巷道中矿工的监控图像。为了在低照度的巷道监控图像中分割矿工的脸部,研究了在YCbCr颜色模型的各分量中矿工脸部肤色的统计特征;在Cr分量空间,采用统计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邻域平滑处理和再次分割,大幅度降低背景像素的影响,缩小脸部区域的范围。试验结果表明,该方法计算简单、速度快,能成功地分割巷道监控图像中的脸部区域。

It is very important in practice that miners＇ face segmentation which can benefit indexing miners＇ images of coal mine surveillance images files. The miners＇ surveilled images are sampled at typical monitorring angles in laneway. In order to segment miners＇ faces in dim-lightening laneway monitor images, the statistic feature of miners＇ face color is analyzed separately in tri-component of YCbCr color model. The pixels of skin color area and the background area that seem to shin color area are segmented in Cr component based on the segmentation method with statistic 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 again. The results of experiments show that the method above are simple and celerity in computing and can segment the faces area in laneway monitor images successfully.