同步轨道目标的单帧检测方法

Single Frame Detecting Method of Geostationary Arti?cial Objects

摘要:

光电望远镜监测同步轨道目标时，一般采用凝视目标和凝视恒星两种工作模式，在这两种工作模式下，由于恒星和目标相对地球运动速度不同，它们在CCD上成像会有所差别，即圆形小目标和短划线。文章针对它们成像的差异，提出了一种利用目标几何形状来区分目标和恒星的方法，该方法先利用形态滤波对图像进行预处理，提高目标信噪比的同时又不破坏目标的形状特征，然后通过二值分割、特征标记计算出目标的紧凑度、通过目标的原点矩计算出目标外接椭圆的长短轴，最后将目标的紧凑度和外接椭圆短长轴比进行特征融合，对融合结果选择合适门限来区分目标和恒星。试验结果表明，所提出的这种特征融合检测方法是一种有效的单帧目标检测方法。

Abstract:

When optical telescope is used for monitoring synchronous orbit targets, gazing targets and gazing stars are two kinds of general working mode. And due to the different speed of stars and targets relative to the Earth, they images on the CCD will be different as round shaped small targets and dashes. For their differences in image, a method is proposed to distinguish targets and stars by using target geometric shape, in which, firstly morphological filter is used to make image pre-processing to improve signal noise ratio of targets without disturbing the shape features of targets. Then the compactness of targets is calculated by binary segmentation and the feature marks, and the external ellipse’s major and minor axis is calculated based on origin moment. Finally, the compactness of targets and the minor to major axial ratio of the external ellipse are feature fused, and the suitable threshold is chosen from fusion results to distinguish between targets and stars. The results show that the proposed method based on feature fusion is an effective single-frame target detection method.