基于PolSAR图像的机场跑道与飞机目标检测

发布时间：2018-01-26 07:40

本文关键词： 极化合成孔径雷达极化散射特征极化目标分解机场跑道检测飞机检测　出处：《中国民航大学》2014年硕士论文　论文类型：学位论文

【摘要】：极化合成孔径雷达(Polarimetric Synthetic Aperture Radar,PolSAR)可对感兴趣区域(Region of Interest,ROI)实施全天时、全天候的侦察。同时,通过不同的天线组合方式获取反映目标振幅、相位等信息的极化散射矩阵,从而比传统的单极化SAR获得更为丰富的地物信息。已被广泛应用于农业、林业、地质学、海洋学、军事探测等众多领域。机场作为军用和民用的重要设施,其自动检测在军事侦察、精确打击、紧急救援和飞机导航等众多领域有着重要的实用价值。飞机作为一种典型的人造目标,是军事侦察的主要打击目标之一,针对飞机的检测研究具有重要的意义。本文针对复杂场景下的PolSAR图像机场跑道检测和飞机目标检测分别展开研究。对于机场跑道检测,本文在研究了PolSAR图像分类和机场跑道特征的基础上,给出了两种复杂场景下的机场跑道检测算法。第一种算法首先利用先验信息结合H/α分类提取模板;然后利用PolSAR图像极化相干矩阵的统计特性进行分类;最后利用跑道尺寸和结构特征进行判别,确定机场跑道区域。第二种算法作为改进算法,采用h/q分类提取地物模板,并加入极化总功率检测器判别跑道。利用美国UAVSAR系统采集的多组全极化实测数据对两算法进行实验,结果表明,后一种算法在继承了前一种算法能正确检测出跑道的优点的同时,降低了运算量,虚警更少,跑道轮廓更清晰,细节保持更好。从两个方面对PolSAR图像中飞机目标检测展开了研究。一方面,针对复杂大场景,根据飞机通常停靠在机场停机坪、滑行道等区域的特点,给出一种基于先验知识的检测算法。该算法在检测出机场跑道区域的基础上,利用Shannon熵对飞机和跑道加以判别。另一方面,针对复杂小场景,在已经定位了机场区域的前提下,给出一种基于条件熵和Shannon熵的检测算法。该算法利用条件熵和Shannon熵以及飞机的结构特征进行多目标检测。采用美国UAVSAR系统、AIRSAR系统采集的多组全极化实测数据对两算法进行实验,并验证了该算法的有效性。
[Abstract]:Polarimetric Synthetic Aperture Radar. PolSAR can carry out round-the-clock reconnaissance on the area of interest region of InterestROI. At the same time. The polarimetric scattering matrix reflecting the amplitude and phase of the target is obtained by different antenna combinations, which is more abundant than the traditional single-polarization SAR. It has been widely used in agriculture and forestry. Geology oceanography military exploration and many other fields. Airport as an important military and civilian facilities its automatic detection in military reconnaissance accurate strike. As a typical artificial target, aircraft is one of the main targets of military reconnaissance. It is of great significance to study the detection of aircraft. In this paper, the airport runway detection and aircraft target detection based on PolSAR images under complex scenes are studied, respectively. In this paper, PolSAR image classification and airport runway features are studied. In this paper, two algorithms for airport runway detection in complex scenarios are presented. Firstly, the template is extracted by using prior information and H / 伪 classification. Then the statistical properties of polarimetric coherence matrix of PolSAR image are used to classify. Finally, the size and structure of the runway are used to determine the airport runway area. The second algorithm is used as the improved algorithm to extract the ground object template by using h / Q classification. The two algorithms are tested by using multi-sets of fully polarimetric measured data collected by American UAVSAR system, and the results show that the proposed algorithm can be used to identify the runway with a polarimetric total power detector. The latter algorithm not only inherits the advantages of the former algorithm, but also reduces the amount of computation, less false alarm and clearer contour of the runway. The study of aircraft target detection in PolSAR image is carried out from two aspects. On the one hand, according to the complex large scene, the aircraft is usually parked at the airport apron. Based on the characteristics of taxiway and other areas, a priori knowledge based detection algorithm is presented. On the basis of detecting the airport runway area, Shannon entropy is used to distinguish the aircraft from the runway. For complex small scenarios, the airport area has been located under the premise. This paper presents a detection algorithm based on conditional entropy and Shannon entropy. The algorithm uses conditional entropy and Shannon entropy as well as structural features of aircraft to detect multi-target. American UAVSAR system is adopted. . The two algorithms are tested by multi-sets of full polarization data collected by AIRSAR system and the validity of the algorithm is verified.
【学位授予单位】：中国民航大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：V351;TN957.52

【参考文献】