在密集杂波和复杂多目标情况下,雷达自动检测与跟踪产生的虚假航迹较多。针对这一问题,首先利用迭代式阈值分割法从单帧雷达视频图像中快速检测出疑似目标,并对连续多帧检测结果进行积累;然后综合目标的单帧检测结果和多帧积累结果提取目标的面积变化比和能量密度特征,并分别构建二者的基本概率赋值函数;最后,采用D-S证据理论对其进行决策融合,获取最优决策以实现运动目标的自动检测。利用某型导航雷达实测数据验证了算法的有效性。
In order to solve the problem of high false track rate of radar automatic detection and tracking in dense clutter and complex multi-target situation,the suspected targets from a single-frame radar image are firstly detected using the iterative threshold segmentation method,and the accumulation results of the singleframe detection results are gained.Then,the area change ratio and the energy density of targets are extracted based on single-frame detection results and multi-frame accumulation results,and their basic probability assignment functions are built respectively.Finally,the optimal threshold is obtained using decision fusion based on the D-S evidence theory to realize the automatic detection of moving targets.The effectiveness of the algorithm is verified by the collected marine radar data.