首页 >  , Vol. , Issue () : -

摘要

全文摘要次数: 142 全文下载次数: 211
引用本文:

DOI:

10.11834/jrs.20219336

收稿日期:

2019-09-20

修改日期:

2020-09-24

PDF Free   EndNote   BibTeX
一种结合置信度的多通道星载SAR在轨舰船低虚警快速检测方法(人工智能专刊)
郑儒楠, 仇晓兰
中国科学院电子学研究所苏州研究院 苏州
摘要:

舰船目标快速检测是星载合成孔径雷达的重要应用之一,也是SAR星上处理首先发展的应用方向。目前SAR舰船快速检测普遍使用的图像恒虚警(CFAR)检测方法中,低虚警一直是研究的重点,多通道SAR中舰船目标存在多通道虚假目标,进一步增加了虚警剔除的难度。为此,本文提出了一种基于置信度计算和SVM判别的低虚警CFAR舰船快速检测方法。首先,在传统CFAR算法中引入分块计算,同时针对目前舰船的长宽、面积等几何参数统计,设计舰船目标置信度计算方法,并对几何参数相近的虚假目标利用SVM分类器进行舰船目标与虚假目标的快速区分,并设计了CFAR检测总体流程。最后,采用GF-3卫星双通道模式实测数据对已有方法和新方法进行了算法运行时间及目标检测性能对比。实验结果表明,在基于TX2构建的星上处理平台上,新方法运行时间比传统CFAR方法缩短30倍,具有良好的实时性,同时对舰船目标检测检测率达到97.8%的同时,虚警率控制在5.2%,具有良好的虚假目标剔除效果。#$NL

Fast detection method for low false alarm of multi-channel spaceborne SAR image combined with confidence calculation
Abstract:

The rapid detection of ship targets is one of the important applications of spaceborne synthetic aperture radar, and it is also the first application direction of SAR satellite processing. In the current image false alarm (CFAR) detection method commonly used in SAR ship rapid detection, low false alarm has always been the focus of research. Multi-channel SAR targets have multi-channel false targets, which further increases the difficulty of false alarm elimination. To this end, this paper proposes a fast detection method for low false alarm CFAR ships based on confidence calculation and SVM discrimination. Firstly, the block calculation is introduced in the traditional CFAR algorithm. At the same time, the ship"s target confidence calculation method is designed for the current geometric parameters such as the length, width and area of ??the ship, and the SVM classifier is used for the false target with similar geometric parameters. The rapid distinction between ship targets and false targets, and the overall process of CFAR testing was designed. Finally, the GF-3 satellite dual-channel mode measured data is used to compare the algorithm running time and target detection performance between the existing method and the new method. The experimental results show that on the on-board processing platform based on TX2, the running time of the new method is 30 times shorter than that of the traditional CFAR method, and it has good real-time performance. At the same time, the detection rate of the ship target detection is 97.8%, and the false alarm rate. Controlled at 5.2%, with good false target rejection.

本文暂时没有被引用!

欢迎关注学报微信

遥感学报交流群 分享按钮