针对卫星在轨处理图像以获取高价值目标信息的需求, 分析了卫星在轨探测系统技术要求, 开发了由高性能处理器组成的嵌入式系统, 能支持目标探测算法的高速计算。研究了通过中值滤波、阈值分割、连通域标记和角度测量实现目标实时探测的算法, 使用多类样本图像对算法进行系统测试, 计算时间小于1 s, 能够检测出目标并计算其位置信息, 满足在轨探测系统要求。

Aiming at the requirement of satellite on-orbit processing images to obtain high value target information, the technical requirements of satellite on-orbit detection system are analyzed, and the embedded system composed of high performance processors is developed. The system supports high-speed computing for target detection algorithms. The target real-time detection algorithm by median filtering, threshold segmentation, connected components labeling and angle measurement is studied. Using multiple types of sample images, the test results indicate that the calculation time is less than one second, the target can be detected and its position can be computed, and on-orbit detection system requirements are met.