基于形态学的白天近红外星目标检测

受多种因素的影响,白天获得的近红外星图像信噪比低,且背景通常是不均匀的,为红外星目标的提取造成了极大的困难。通过理论计算,实际白天拍摄的近红外星图信噪比极低,使用一般的滤波方法无法提取出恒星目标,形态学运算对于星图像的处理具有较好的效果。首先分析目标及噪声大小,选择较大的结构元素,然后组合开运算及顶帽变换,提取出背景,得到只含有目标及噪声的图像,然后改变结构元素大小,消除掉噪声,进行目标分割。实验证明,可较好地分离出恒星目标。

Near-infrared Star Target Detection Based on Morphology during Daytime

Influenced by many factors, the near-infrared star images have a low SNR. And the background is always non-uniform, it brings great difficulties in the process of picking up near-infrared star target. According to theoretical calculation, near-infrared star images taken in actual daytime have a very low SNR, star targets can not be picked up by common filtering method. Morphology algorithm is excellent to deal with star images. Firstly, the size of targets and noise are analyzed, the structure elements with bigger size are chosen. Based on assembled opening operation and top-hat transformation, the background is picked up and the images which only have targets and noise are got. Then the sizes of structure elements are changed in order to remove the noise and segment the targets. The experiment shows that star target can be perfectly separated.