传统基于统计模型的雷达目标检测算法往往仅利用雷达探测回波的能量信息分辨雷达目标,其难以适应密集杂波环境下的雷达探测感知场景,从而显著降低雷达目标探测性能。针对密集杂波背景下的雷达目标检测难题,本文引入随机森林算法,提出一种基于随机森林的雷达目标多维特征检测方法。利用随机森林方法充分融合雷达探测目标回波中时 间维、空间维、距离维等多个维度的特征,形成对雷达目标与环境杂波差异的深度刻画,从而实现精准鲁棒的雷达目标检测,显著提升雷达的目标检测性能。最终,基于雷达实测数据对本文所提方法的性能进行验证,实验结果充分验证了本文所提算法的有效性。

Traditional methods based on statistical model only utilize the energy information, which could not adapt to the highly complex and dynamic battlefield environment and result in significantly decrease of Radar detecting performance. Concerning this issue, Random Forest is introduced to take fully advantage of multi-dimensional feature for Radar target detection in this paper. By utilizing the multi-dimensional features of distance, time and space dimension of the target, the difference between radar target and clutter is deeply expressed. Thus it improves target detection performance of Radar under strong clutter significantly. Finally, the performance of the proposed method is verified based on experimental measured data, and the experimental results validate the effectiveness of the proposed method.

TN957. 51