一种基于改进霍夫变换的天基GEO目标快速检测算法
2018-11-13翟永立丁雷裴浩东
翟永立 丁雷 裴浩东
摘 要: 分析GEO空间目标和恒星在高轨探测器上的运动特性和成像特点,提出一种基于改进Hough变换的空间目标检测算法,以便有效地从复杂星空背景下提取出空间目标。采用均值加方差的方式选取全局阈值;根据末帧差分图的疑似目标点确定进行Hough变换的区域和潜在的轨迹端点;使用改进的单维Hough变换进行轨迹判断。实验结果表明,与传统Hough对比,算法的复杂度、占用内存和计算量都有明显降低,算法所用时间减小50%,可见该算法可以用于空间目标在轨实时识别,并且对噪声有不错的抵抗能力。
关键词: GEO空间目标; 在轨识别; 目标检测; 阈值分割; 复杂星空背景; Hough变换
中图分类号: TN957.52?34; TP391.9 文獻标识码: A 文章编号: 1004?373X(2018)22?0001?04
Abstract: A space target detection algorithm based on the improved Hough transform is proposed by analyzing the motion feature and imaging characteristics of the GEO space targets and stars on the high?orbit detector, so as to effectively extract space targets from the complex celestial background. The global threshold is selected by using the mean value and variance. The region for the Hough transform and the potential trajectory endpoints are determined according to the suspected target points of the last frame difference image. The improved single?dimensional Hough transform is used to judge the trajectory. The experimental results show that, in comparison with the traditional Hough, the algorithm has obviously?reduced complexity, occupied memory and computation quantity with its time consumption reduced by 50%, which indicates that the algorithm can be used for real?time on?orbit recognition of space targets, and has a good anti?noise capability.
Keywords: GEO space target; on?orbit recognition; target detection; threshold segmentation; complex celestial background; Hough transform
0 引 言
空间目标泛指人造地球卫星、空间碎片等地球附近的快速运动天体。空间目标的监视对空间态势的感知、保障在轨运行航天器的安全有着重要意义。而地球静止轨道(GEO)因其轨道特殊性,常常运行着各国天基信息网的关键节点卫星[1],所以对地球静止轨道带内空间目标的跟踪观测便显得极为重要,如美国2016年发射的gssap卫星[2]。
空间目标距离探测器较远时,其所成像仅占据有限个像素,为点状分布[3]。星图中恒星数量巨大,目标点完全淹没在其中。如何克服恒星及噪声的影响,对空间目标运动轨迹进行精确提取,是当前国内外航天科技领域研究的难点。
针对从序列图像中检测出运动点目标,国内外学者提出了许多检测算法,如光流法、匹配滤波和动态规划法等。……
