【作者】 刘银中；

【导师】 李绍滨；

【作者基本信息】 哈尔滨工业大学 ， 信息与通信工程， 2006， 硕士

【摘要】 机载合成孔径雷达是一种以飞行器为平台,工作在微波波段的主动式遥感器。由于它具有全天候、全天时观测目标的二维高分辨图像的突出优点,所以在许多方面都获得了广泛应用。本文主要根据机载合成孔径雷达实测数据进行多普勒参数估计和成像算法的研究。本文首先回顾了机载合成孔径雷达的一些基本问题,介绍了合成孔径雷达的基本原理、距离向和方位向分辨率、距离徙动、驻定相位原理等为后续章节做了理论上的铺垫。场景成像的合成孔径雷达一般是正侧视放置,但是载机不可避免的要受到各种扰动影响,雷达会有一定的斜视角,从而导致了多普勒中心频率不为零,多普勒调频斜率也随着雷达斜视角改变。所以为了方位向更好的聚焦,我们需要从实测数据中估计多普勒参数。在多普勒参数估计中,采用了最大值法、能量法和自相关函数法进行多普勒中心频率估计;采用了图像偏移法、时频分析法、最小熵法和分数阶傅里叶变换法进行多普勒调频斜率估计。在多普勒中心频率估计中,最大值法简单,但精度不高,自相关函数法有较高的估计精度。在多普勒调频斜率估计中,时频分析法比较直观,估计精度也不错,脱机处理中获得了广泛的应用,图像偏移法估计速度快,在运动补偿中得到了应用。分数阶傅里叶变换的估计精度不是很高,但是这种信号处理的方法还有很多待开发的作用。根据估计得到的多普勒中心频率和多普勒调频斜率我们计算出载机的速度和雷达天线相位中心斜视角,从而进行方位向精确聚焦。接下来,本文实现了传统的RD算法、几种改进的RD算法、CS算法和RM算法,并对算法的实现过程给与详细讨论。最后结合估计的多普勒参数对机载合成孔径雷达实测数据进行成像处理,从图像结果可以看出CS比RD算法要精确些,RM算法是严格匹配算法,但需要插值,计算量较大,插值精度对成像质量有较大影响。最后将实测数据单视与多视成像结果进行比较,可见多视处理可以显著的抑制相干斑噪声。更多还原

【Abstract】 Airborne Synthetic Aperture Radar (SAR) is a kind of active remote sensor, which works in the microwave band and takes aircraft as the working platform. SAR has the whole weather and the whole day characteristics that observing two dimensions high resolution image of the object, now which has widely been applied in many fields. This paper mainly investigated the estimation methods of doppler parameters and the imaging algorithms of airborne stripe SAR.First, we go over the basic theory of SAR making a theoretical preparation for following chapters, introducing the principle of SAR, range and azimuth resolution ,range migration, principle of the stationary phase. The radar which is embedded on platform sidelooking, but the platform will be disturbed by airflow inevitably. So, the radar will have a small sloping angle, thereby, the doppler centroid that is not zero and doppler frequency rate need to be estimated from the raw echo data in order to focus in azimuth commendably. In the following parts, the paper gives some methods to estimate doppler parameters. At the doppler centroid aspect, this paper make use of Maximum value method; Energy method and autocorrelation function method to estimate doppler centroid. At the doppler frequency aspect, this paper make use of Image offset method; Time-frequency method; Minimum entropy and fractional fourier transformation method to estimate doppler frequency rate. Maximum value method is fast but not accurate, autocorrelation function method is accurate and have a good effect form radar image according to their estimate methods. Image offset method that is also fast can be applied to motion compensation, time-frequency method is intuitionistic and have a better accurate, which is applied extensively on offline. Fractional fourier transformation method is not accurate, but which have some now application in signal processing. We can obtain the velocity of aircraft and the angle of radar antenna phase center in order to focus in azimuth accurately.Finally, this paper actualize RD algorithm, improved RD algorithm, CS algorithm, RM algorithm, and then processing raw echo data make use of some algorithms that have been introduced and adopt doppler parameter that have been estimated above. From radar images, we can find that CS algorithm is more更多还原