The exploitation of multi-look synthetic aperture radar and inverse synthetic aperture radar images for non-cooperative target recognition
The exploitation of multi-look synthetic aperture radar and inverse synthetic aperture radar images for non-cooperative target recognition
复制标题
利用多视合成孔径雷达和逆合成孔径雷达图像进行非合作目标识别
DOI：
--
复制
发表时间：
2007
期刊：
影响因子：
--
通讯作者：
S. Papson
中科院分区：
文献类型：
--
作者：
S. Papson
合成孔径雷达（SAR）和逆合成孔径雷达（ISAR）在非合作目标识别（NCTR）方面的应用已经得到了验证。两种传感方式都能够以稳健的方式提供操作信息。随着战场上处理能力和通信能力的提高，利用SAR/ISAR系统的新机会出现了。同一目标的多个外观现在可以用来提高性能，并允许系统在最大范围内操作。在这项工作中提出的研究概述了在多种外观存在下利用SAR/ISAR图像的各种方法。具体成就包括：开发了一种从SAR图像中提取信息的自动分割方法；开发新的图像融合规则，以整合来自少量独立SAR/ISAR系统的数据；开发持久性框架，以增强大型方面变化数据集中的目标特征；发展了基于SAR图像中多个目标的阴影特征对其进行分类的阴影分类技术；并进行了操作分析，以确定算法在现实场景中的表现。这些算法在各种规范和现实世界的数据集上进行了测试。
Synthetic aperture radar (SAR) and inverse synthetic aperture radar (ISAR) have proven capabilities for non-cooperative target recognition (NCTR) applications. Both sensing modalities have been able to provide operational information in a robust manner. As processing power and communication capabilities on the battlefield increase, new opportunities for exploiting SAR/ISAR systems emerge. Multiple looks of the same target can now be used to increase performance and allow systems to operate at maximum ranges. The research presented in this work outlines a variety of methods for utilizing SAR/ISAR images in the presence of multiple looks. Specific accomplishments include: the development of an automated segmentation method to extract information from SAR images; the development of novel image fusion rules to integrate data from small numbers of independent SAR/ISAR systems; the development of a persistence framework to enhance target features in large, aspect-varying datasets; the development of a shadow classification technique to classify multiple targets based only on their shadow features in SAR imagery; and operational analyses were performed to determine how the algorithms would perform in realistic scenarios. The algorithms are tested on a variety of canonical and real-world datasets.