基于最大值投影和快速配准的空间小目标检测

天基观测平台下弱小目标的检测是分析空间安全的重要研究内容。由于空间中存在大量外观与目标相似的恒星导致可利用空间分布信息缺乏；观测平台的不规则性运动导致帧间成像差异，都使得开发自动快速处理算法的难度增加。在分析星空图像模型的基础上，提出了一种基于三角形匹配和最大值投影的小目标检测方法。首先通过特征三角形对序列图像进行配准，并采用星点坐标矩阵的方法减小计算量。然后针对序列帧所有图像，采用最大值投影变换的方法，检测运动的小目标。最后通过200帧观测图像对算法进行验证，实验表明该方法能实时、准确地对目标进行检测，同时可以精确地定位目标质心。

Space small targets detection based on maximum projection and quick registration

It is an important research content for analysis of space security to detect the small targets with spac-based observation platform. But it's difficult to develop a fast, robust and automatically processing algorithm for some reasons:there are too many stars which have similar appearance with targets in space image and the observation platform moves irregularly. Based on analyzing star image model, a method of space target detection was proposed by triangle matching and maximum projection. First step is the sequence images registration with feature triangle and for reducing computation, the star centroid coordinate matrix was calculated. Then the maximum projection transformation for image sequences was used for detecting the moving small targets. Finally, simulation and test demonstrate the method can achieve real time and accurate targets detection with 200 sequence images.