SAR图像舰船目标探测及其样本制作

发布时间：2018-04-27 00:31

本文选题：合成孔径雷达 + 舰船检测　；参考：《合肥工业大学》2014年硕士论文

【摘要】：海洋舰船目标监视监测是我国一项传统的国防任务。合成孔径雷达(Synthetic Aperture Radar, SAR)是一种较高分辨率成像雷达,具有全天时和全天候的工作能力。基于SAR的舰船检测一直是目标检测重要的研究内容之一。SAR图像舰船探测本质是从海杂波中将舰船目标分割出来,以及目标特征提取与类型识别。然而随着SAR技术的不断发展和对舰船检测要求的进一步提高,一系列新的命题不断产生。基于迫切的实际需求,本文展开了如下研究工作：第一,研究了SAR舰船检测背景和意义,比较了国内外发展现状,总结出国外基于应用角度发展舰船检测的情况,以及国内专注舰船检测核心算法的研究现状；分析了舰船检测亟待解决的新的问题和方向。第二,探究了在舰船检测应用方面SAR图像成像机理,分析并给出舰船检测中满足实际需求的SAR图像预处理常用方法,如杂波滤除、入射角校正和几何校正；分析了影响舰船检测的SAR系统因素和海况因素等。第三,归纳了基于恒虚警率(Constant False Alarm Rate, CFAR)舰船检测方法中海杂波统计经典模型,学习了不同统计模型的公式推导、参数选择和适用条件,并比较了概率模型优缺点等；并在此基础上介绍了目前研究较多的多阶CFAR算法。第四,针对目前国内在舰船识别方向虽然理论分析较多,多数基于人工经验和仿真数据进行的,为了分析SAR舰船目标检测算法性能和开展进一步研究,详细介绍了基于实测舰船信息分析的实验和基于SAR图像和舰船自动识别信息联合探测的实验。基于上述研究学习讨论,围绕舰船检测迫切需要解决的相关问题,开展了本文较为创新的工作,如下：(1) 针对高噪声条件下单视复SAR图像研究了基于Rice分布的CFAR检测算法,改善了传统概率模型在噪声较大情况下检测精度较低的情况。(2)对SAR图像中舰船目标特征分析,并考虑到CFAR算法中虚警率的调设过多地依赖人工先验知识,使得最终检测结果不稳定。研究了基于SAR目标高亮度、高密集度特征的舰船检测算法,并与传统的形态学处理方法进行比较,优势较为明显。(3)详细叙述了高分辨率SAR图像舰船样本库建设问题,并研究了舰船目标几何特征提取方法,在此基础上分析了极化方式和入射角对舰船长度的影响,为开展SAR图像舰船类型识别做了一定的基础。
[Abstract]:The monitoring and monitoring of marine ship targets is a traditional national defense task in China. Synthetic Aperture Radar (SAR) is a high resolution imaging radar. It has all day and all-weather working ability. The ship detection based on SAR is one of the important research contents of the target detection. The essence of the.SAR image ship detection is the essence of the ship detection. The ship targets are segmented from the sea clutter, and the target feature extraction and type recognition are extracted. However, with the continuous development of SAR technology and the further improvement of ship detection requirements, a series of new propositions are constantly produced. Based on the urgent practical needs, the following research work has been carried out in this paper. First, the background of SAR ship detection is studied. And significance, compare the development situation at home and abroad, summarize the situation of developing ship detection based on the application angle in foreign countries, and the research status of the core algorithm focusing on ship detection in China; analyze the new problems and directions to be solved in ship detection. Second, explore the imaging mechanism of SAR image in the application of ship detection and analyze and give it The common methods of SAR image preprocessing, such as clutter filtering, incident angle correction and geometric correction, are produced to meet the actual needs of ship detection, such as the SAR system factors and sea condition factors affecting ship detection. Third, the classical model of the sea clutter based on the Constant False Alarm Rate (CFAR) method is summed up. The formula derivation of different statistical models, parameter selection and application conditions are studied, and the advantages and disadvantages of the probability model are compared. On the basis of this, the multi order CFAR algorithm is introduced. Fourth, although there are many theoretical analyses in the direction of ship recognition in China, most of them are based on artificial experience and simulation data. The performance and further research of SAR ship target detection algorithm are analyzed, and experiments based on measured ship information analysis and joint detection based on SAR image and ship automatic recognition information are introduced in detail. Based on the above research and discussion, the relevant problems which are urgently needed to be solved around ship detection are discussed, and the innovation of this paper is carried out in this paper. The following work is as follows: (1) the CFAR detection algorithm based on Rice distribution is studied for single visual complex SAR images under high noise conditions, which improves the situation of low detection precision in the traditional probability model with large noise. (2) analysis of ship target characteristics in the SAR image, and considering the excessive dependence of false alarm rate in the CFAR algorithm. The test results are unstable. The ship detection algorithm based on the high brightness and high density characteristics of the SAR target is studied. Compared with the traditional morphological processing method, the superiority is more obvious. (3) the construction of the high resolution SAR image ship sample database is described in detail, and the geometric feature extraction side of the ship target is studied. On this basis, the influence of polarization mode and incidence angle on ship length is analyzed, which provides a basis for ship type recognition in SAR images.

【学位授予单位】：合肥工业大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TN957.52

【相似文献】