基于逆合成孔径雷达的空中目标的三维重构算法研究

发布时间：2018-06-04 03:22

本文选题：逆合成孔径雷达 + 平动补偿　；参考：《哈尔滨工业大学》2014年硕士论文

【摘要】：逆合成孔径雷达(Inverse Synthetic Aperture Radar,ISAR)实现了对一些观测目标如飞机、舰船、导弹等进行成像和识别的功能,近年来在各领域得到了广泛的应用。早期的雷达成像是将物体以“点”的形式来进行检测的,二维成像使人们突破了早期雷达以“点”目标形式观测远处物体的限制,但是这种成像实质是目标的三维结构在二维平面上的投影,存在收集信息不全面的弊端。随着科学技术的飞速发展和科研工作的不断深入,人们对雷达成像的精度要求也越来越高,为了能够从雷达回波中提取更全面的目标信息,三维成像已成为当前热门的研究趋势,三维重构技术是实现三维ISAR成像的主要途径之一,本课题对于基于ISAR的三维重构技术的三维成像研究具有重要的理论和应用意义。论文主要是在单天线的雷达系统基础上进行研究的,利用获得的不同姿态下的目标的二维ISAR图来进行三维重构研究,采用因子分解法来进行重构。论文完成的工作主要包括,对观测目标实现ISAR二维成像,特征点的提取和匹配,近地轨道模型的横向定标,目标的三维重构等。论文对于ISAR转台模型设计了系列基本模型,比如规则的图形和飞机模型,针对这些模型利用成像技术得到二维图。为了得到重构需要的二维坐标信息,需要对不同时刻的二维图进行特征点的提取。考虑到目标的不断变化,为了保证对不同时刻的ISAR二维图提取出来的特征点能够匹配上,需要对特征点的匹配算法进行深入的研究。论文主要介绍了空间极值和Harris角点检测两种特征点提取的算法。得到特征点后采用尺度不变转换方法(SIFT)算法进行匹配,该方法是基于多尺度空间计算提取出的特征点的特征向量,将不同帧的关键点的特征向量都计算出来,根据特征向量进行特征点的匹配。上述工作完成后,得到了不同时刻的目标的二维图坐标信息,再利用因子分解方法得到三维重构图。已有的研究针对ISAR二维图的特征点的匹配主要是利用网格法和光流场匹配追踪法等技术来完成,论文主要利用上面提到的角点提取和匹配的方法进行匹配追踪,然后利用因子分解方法进行三维重构。最后对设计的模型和实测的空间站数据进行三维重构,验证了特征点提取和匹配追踪方法以及三维重构方法的有效性,同时也证明了在更加精确的匹配前提下的重构效果也更好。
[Abstract]:Inverse Synthetic Aperture radar (ISAR) has been widely used in recent years for imaging and recognition of some observation targets such as aircraft, ships, missiles and so on. Early radar imaging was used to detect objects in the form of "dots". Two-dimensional imaging made people break through the limitation of early radar to observe distant objects in the form of "points". But this kind of imaging essentially is the projection of the object's three-dimensional structure on the two-dimensional plane, which has the disadvantage of collecting incomplete information. With the rapid development of science and technology and the deepening of scientific research, the accuracy of radar imaging is becoming more and more high. In order to extract more comprehensive target information from radar echo, 3D imaging has become a hot research trend, 3D reconstruction technology is one of the main ways to realize 3D ISAR imaging. This topic has important theoretical and application significance for 3D imaging research based on ISAR. This paper is mainly based on the radar system of single antenna. The 3D reconstruction is carried out by using the two-dimensional ISAR diagram of the target with different attitude, and the factor decomposition method is used to reconstruct the target. The work accomplished in this paper mainly includes ISAR 2D imaging, feature point extraction and matching, lateral calibration of Leo model, 3D reconstruction of the target, and so on. In this paper, a series of basic models are designed for ISAR turntable models, such as regular graphics and aircraft models. In order to obtain the 2D coordinate information needed for reconstruction, it is necessary to extract the feature points of 2D images at different times. Considering the continuous change of target, in order to ensure that the feature points extracted from two-dimensional ISAR images at different times can match, it is necessary to study the matching algorithm of feature points in depth. This paper mainly introduces two feature extraction algorithms: spatial extremum and Harris corner detection. After getting the feature points, the scale invariant transformation method (sift) is used to match the feature points. The method is based on the multi-scale space to calculate the feature vectors of the feature points, and calculate the feature vectors of the key points in different frames. The feature points are matched according to the feature vectors. After the above work is completed, the coordinate information of the two-dimensional graph of the target at different times is obtained, and then the three-dimensional reconstruction graph is obtained by using the factorization method. The existing research is mainly based on mesh method and optical flow field matching method to match the feature points of ISAR two-dimensional images. The paper mainly uses the corner extraction and matching method mentioned above to match and track. Then the factor decomposition method is used to reconstruct three-dimensional. Finally, the 3D reconstruction of the designed model and the measured data of the space station is carried out to verify the effectiveness of the feature point extraction, matching and tracking methods, and the 3D reconstruction method. At the same time, it also proves that the reconstruction effect is better under the premise of more accurate matching.
【学位授予单位】：哈尔滨工业大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TN957.52

【共引文献】