The CCD imaging sensor has the inherent characteristics of frame transfer. When bright stars appear in the CCD sensor, a bright line will be produced in its row or column. Smear phenomenon changes the nature of the coverage area of the image, so it must be removed. On the analysis of the mechanism of the Smear, a Smear removal algorithm is proposed, which can remove the bright lines and reserve dim targets. According to local histogram Gaussian fitting, the image background parameters are estimated, and the median filter and the multi-threshold are used to reserve the dim targets. In order to improve the visual quality, a sub-sampling method and a 4-neighbour domain filter are proposed. Experiment verifies the effectiveness of the method.%星空中的 CCD 图像由于帧转移的固有特性，当星图中有亮星出现时，会在亮星所在的行或列产生一条亮线，称为Smear效应，使覆盖区域的图像特性发生改变并对目标检测造成干扰，必须对其进行消除。本文在对Smear产生机理的分析上，提出一种有效去除亮线并保留弱小目标的方法。首先通过局部直方图Gauss拟合的方法对背景参数进行估计，利用中值滤波和多阈值的设定保留了弱小目标。最后通过亚采样方法对目标进行增强，四邻域滤波对图像进行修复，提高了图像的视觉质量。通过实验验证了该方法的有效性。
展开▼
