SAR图像舰船检测方法研究

发布时间：2017-12-28 07:26

本文关键词：SAR图像舰船检测方法研究　出处：《沈阳航空航天大学》2017年硕士论文　论文类型：学位论文

【摘要】：SAR舰船目标检测在军事、渔业等领域有大量应用,开展SAR图像舰船目标检测具有重要意义。本文的主要研究内容是SAR图像的海陆分割和舰船目标检测两个方面。本文首先分析了SAR的成像机理和图像特性,噪声产生原因和特点。介绍了传统检测方法,并分别从海杂波统计特性、影响舰船检测的因素两个方面对检测方法做了分析,提出了舰船目标检测的方法流程。精确的海陆分割是舰船目标检测的前提和保障,本文提出了利用海洋和陆地在边缘信息和灰度信息上的差异,结合区域生长等的海陆分割方法。对SAR图像做边缘检测时,基于比值的ROEWA算子无法确定边缘的方向,本文首先对图像取对数,将乘性噪声模型转化为加性噪声模型,然后借鉴了Canny算子的思想方法,求取边缘方向,再利用海洋和陆地边缘和灰度的差异,在海域中选择区域生长点,通过区域生长实现海陆分割。在检测舰船时,本文针对传统双参数CFAR检测算法在检测弱小目标时出现漏警,以及遍历图像时出现目标不能被完全检测出来的问题,提出了改进的CFAR检测算法,在做目标分割时,结合图像灰度直方图,提出了针对每个窗口的自适应阈值。本文首先确定舰船可能存在的位置,再对每个检测窗口做阈值分割,提取出舰船。实验结果表明,采用本文提出的对ROEWA算子的改进方法能够很好的检测SAR图像边缘,并较好的完成海陆分割,在目标检测阶段本文改进的CFAR检测算法能够有效降低虚警和漏检,本文提出的自适应检测阈值,在做目标分割时,比Otsu方法对噪声的敏感度低,鲁棒性好。
[Abstract]:SAR ship target detection has a large number of applications in military, fishery and other fields. It is of great significance to carry out SAR image ship target detection. The main contents of this paper are two aspects of sea land segmentation and ship target detection in SAR images. In this paper, the imaging mechanism and image characteristics of SAR, the cause and characteristics of noise generation are analyzed. The traditional detection methods are introduced, and the detection methods are analyzed from two aspects: the statistical characteristics of sea clutter and the factors affecting the detection of ships. The flow chart of ship target detection is also presented. Accurate land and sea segmentation is the premise and guarantee for ship target detection. This paper proposes a land sea segmentation method based on the difference between ocean and land in edge information and gray information, combined with regional growth. Edge detection of SAR image, ROEWA operator edge ratio cannot be determined based on the direction based on the logarithmic image, the multiplicative noise model into additive noise model, and uses the method of the Canny operator to get the edge direction, using the difference between marine and terrestrial edge and gray scale, selection regional growth point in the waters, realize the sea land segmentation by Region growing. In ship detection, according to the traditional double parameter CFAR detection algorithm in small targets detection when the leakage of police, as well as problems can not traverse the image target is completely detected, put forward improved CFAR detection algorithm, in object segmentation, combining the gray histogram of image, the adaptive threshold for each window is proposed. In this paper, the possible location of the ship is determined first, and then the threshold segmentation is made for each detection window, and the ship is extracted. The experimental results show that the improved method of ROEWA operator is proposed in this paper can detect the edge of SAR image is very good, and better to complete the land segmentation, can effectively reduce the false alarm and missing in the CFAR detection algorithm of the improved target detection stage, adaptive threshold is presented in this paper, in object segmentation, Otsu method low sensitivity to noise, robustness.
【学位授予单位】：沈阳航空航天大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TN957.52

【相似文献】