基于运动信息的星图中空间目标检测算法

为了对空间目标进行精确定位与跟踪，建立目标运动轨迹，对基于运动信息的星图中空间目标检测算法进行了研究。首先，基于空域目标能量累积的方法提高目标信噪比，增大弱小目标被分割出来的概率。然后，根据相邻两帧星图中恒星相对位置不变性提取星图控制点，并根据控制点坐标求取全局运动参数。然后，根据分割出的星点与控制点的位置关系，将恒星与潜在目标分类。接下来，利用3帧最近邻关联法粗检测目标，并利用多帧前后向搜索法滤除虚假目标。最后将序列星图中所有目标编号，建立起目标的轨迹针对4组实拍星图的实验结果表明：所有目标的运动轨迹全部建立，检测到目标的最小平均信噪比为2.99，目标最小平均运动速度2.47 pixel/frame，最大平均运动速度12.72 pixel/frame。本文算法基本满足星图中空间目标检测的检测概率高、虚警少和轨迹定位精度高等要求。

Space target detection in star image based on motion information

In order to precisely locate and track space targets, and construct targets’ kinematic trajectories, a space target detection algorithm in star image based on motion information is researched. Firstly, target’s SNR is increased based on spatial energy accumulation, and the faint targets can be easily segmented from the background. Then, according to the invariance of stars’ relative position in two neighboring frames, the control points of star images are extracted. And the global motion parameters can be calculated with the control points in succession. Then, according to the relative position between star points and the control points, stars and potential targets are classified. And then, targets are grossly detected utilizing the 3-frames nearest neighboring correlation method, and false targets are filtered with the multiframe back and forth searching method. In the end, all targets in star image sequence are numbered, and targets’ trajectories are constructed. Experimental results about four groups of real photographed star images are as follows: All targets’ kinematic trajectories are constructed. The average SNR detected is 2.99. The least mean motion velocity is 2.47 pixel/frame. The most mean motion velocity is 12.72 pixel/frame. The algorithm in this paper can satisfy the space target detection requirements, which include high detection probability, few false alarms and high trajectory locating accuracy, etc.