基于Wishart分类的极化SAR图像机场跑道区域检测

发布时间：2018-11-05 12:30

【摘要】：极化合成孔径雷达(Polarimetric Synthetic Aperture Radar,简称PolSAR)是一种先进的、工作在微波频段的成像系统,可以不受天气、光照等因素的约束,全天时全天候地获取地物目标信息。与早期单极化成像雷达相比,极化SAR系统的多通道设计使其能够测量得到多种极化方式组合下的散射回波的幅度、相位等极化信息,识别地物的能力更强。因此,极化SAR在图像分类、目标检测与识别等应用领域中引起了越来越广泛的关注。跑道作为机场的核心设施,其自动检测与识别在无人机和飞机的自主着陆系统、导弹精确打击、紧急救援等方面具有极其重要的应用价值。本文基于地物的极化散射特性,展开了对机场跑道区域检测的研究,主要工作内容如下:1、将地物的极化散射特性和跑道结构特征相结合,给出了一种基于Wishart无监督分类的极化SAR图像机场跑道区域检测方法。首先应用极化散射熵和Freeman分解将PolSAR图像粗分类,并结合Wishart分类器对粗分类结果迭代修正;然后利用机场跑道的弱散射回波特性从分类结果中提取感兴趣区域(Region of Interest,ROI),最后结合机场跑道的结构特征对感兴趣区域进行筛选辨识,确定机场跑道区域的位置。2、针对算法一运算量大的问题,本文研究给出了一种基于超像素的机场跑道区域检测方法。该算法首先使用简单线性迭代聚类方法(Simple Linear Iterative Clustering,SLIC)把PolSAR图像中位置相邻、极化特性相似的像素点聚合成一个大小可控的超像素;然后结合极化散射特性与Wishart分类器对这些超像素分类,并从分类结果中提取ROI;最后把跑道的结构特征应用到ROI筛选辨识中,定位跑道区域位置。对多组极化SAR实测数据进行实验来验证各算法的有效性,实验结果表明:两种检测方法在地物先验信息少的情形下依然能够有效检测机场跑道区域,推广性好。在算法运算量上,算法二运算量远低于算法一,而在边缘细节保持以及结构完整上,算法二性能比第一种方法较差。
[Abstract]:Polarimetric synthetic Aperture Radar (PolSAR) is an advanced imaging system working in microwave frequency band. It can obtain the information of ground objects all day and all day without the constraints of weather and illumination. Compared with the early single-polarization imaging radar, the multi-channel design of the polarimetric SAR system enables it to measure the amplitude and phase of scattering echo in a variety of polarimetric combinations, and has a stronger ability to identify ground objects. Therefore, polarized SAR has attracted more and more attention in image classification, target detection and recognition. Runway as the core facility of the airport, its automatic detection and recognition has extremely important application value in UAV and aircraft autonomous landing system, missile accurate strike, emergency rescue and so on. Based on the polarimetric scattering characteristics of ground objects, this paper studies the detection of airport runway area. The main work is as follows: 1. Combining the polarization scattering characteristics of ground objects with the structural characteristics of the runway. A method of airport runway detection based on Wishart unsupervised classification is presented in this paper. Firstly, the rough classification of PolSAR images is made by polarization scattering entropy and Freeman decomposition, and the result of rough classification is modified by using Wishart classifier. Then the region of interest (Region of Interest,ROI) is extracted from the classification results by using the weak scattering echo characteristics of the airport runway. Finally, the region of interest is selected and identified by combining the structural features of the airport runway, and the location of the airport runway region is determined. In order to solve the problem of large computational complexity, this paper presents a super-pixel based detection method for airport runway area. The algorithm firstly uses simple linear iterative clustering method (Simple Linear Iterative Clustering,SLIC) to aggregate the adjacent pixels in PolSAR images with similar polarization characteristics into a controllable super-pixel. Then combining polarization scattering characteristics with Wishart classifier to classify these super-pixels, ROI; is extracted from the classification results. Finally, the structural features of the runway are applied to the ROI screening identification to locate the location of the runway region. The experimental results show that the two detection methods can effectively detect the airport runway area without prior information of the ground objects, and the results show that the two methods can be used to detect the airport runway area effectively. The computational complexity of the second algorithm is much lower than that of the first one, but the performance of the second algorithm is worse than that of the first method in terms of the edge detail preservation and the integrity of the structure.
【学位授予单位】：中国民航大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TN957.52

【相似文献】