红外监控图像红眼检测与消除

Detection and elimination of red-eye in infrared monitoring images

主动式红外摄像机在夜视监控视频中极易出现红眼效应,在图像中留下明显的亮圆斑。针对此缺陷,提出一种自动红眼检测与消除算法。首先对红外夜视监控图像进行边缘检测和形态学闭运算,形成连通域;然后通过分析亮圆斑的几何特征提出三个限制条件,筛选出红眼所在的连通域,输出红眼掩膜;最后对红眼区域用灰度替换和平滑滤波方法进行消除处理。该方法无需人脸检测,无需利用色彩信息,实验结果表明,红眼消除效果良好,速度快,为红外灰度图像优化提供了有效的方法。

Since red-eye effect is liable to appear in night vision monitoring video of active infrared camera,the obvious bright and round spots will be left in digital images. For this defect,an algorithm which can detect and eliminate red-eye automatically is proposed. The edge detection and morphological closed operation of monitoring image of infrared night vision were conducted to form connecting areas. Then three constraints are put forward by analyzing the geometric features of the bright spots to select connecting areas that the red-eye are locating,and output red eye mask. The red eye areas are eliminated by using the methods of grayscale replacement and smoothing filtering. There is no need to detect human face and use color information for this method. The experimental results show that the proposed method can eliminate red-eye fast and effectively,which provides an effective method for the optimization of infrared grayscale images.