SAR图像质量评估方法研究

发布时间：2018-08-01 08:16

【摘要】：合成孔径雷达（SAR）因其特有的相干成像机理，成像结果存在诸多不同于光学图像的质量问题，如旁瓣噪声、相干斑等，使得图像动态范围小、对比度低、模糊，严重影响SAR图像解译。质量评估过程是用客观参数对SAR图像质量优劣进行评价的过程，是图像解译前的预处理步骤，具有重要的研究价值。本文针对SAR图像质量评估方法进行深入研究，主要内容如下： 1.研究了SAR图像质量评估的基本方法。RADARSAR、ENVISAT、J-ERS等著名的雷达系统都在地面上建有雷达数据质量监测站，从分辨率、旁瓣强度、几何定位精度、辐射分辨率等方面对雷达回波数据质量进行初步评估。本文根据评估方法的不同将这些参数分为点目标质量参数和面目标质量参数，并具体阐述它们的定义及计算方法。针对点目标难以获取，点目标质量参数存在较大计算误差的问题，提出了用复数SAR图像区域相干相关函数评估点目标质量参数的方法。仿真及实测SAR图像数据都证明该方法用于计算SAR图像的分辨率、峰值旁瓣比、积分旁瓣比等质量指标时是有效的。 2.研究了SAR图像散焦模糊的原理及散焦模糊程度评估算法。散焦模糊是导致SAR图像模糊的重要原因之一，主要由调频率估计误差导致。本文从SAR的成像原理出发，分析散焦模糊对成像结果的影响，得到散焦点目标主瓣展宽、旁瓣升高，形成类似光学图像运动模糊结果的结论。基于此，本文以运动模糊参数评估方法为基础寻找有效的SAR图像散焦模糊参数评估算法，，提出了用图像截断频谱的Radon变换评估模糊方向，大窗口中值滤波器快速去噪，用图像显著性区域平均边缘宽度评估散焦模糊程度的算法。文中定义了一个散焦模糊度评估指标UBD，实验证明该指标能够给出与主观评估相一致的结论。 3.研究了SAR图像对比度评估方法及增强算法。图像对比度反映图像的清晰程度，决定图像的信息获取能力。SAR图像对比度评估不仅是图像质量评估的重要组成部分，还能客观评价图像增强算法的优劣。文中阐述了两种基于人眼视觉特性的对比度评估参数，EME和TEN，实验验证了用EME-TEN参数组合评估SAR图像对比度的有效性。本文还深入研究了将模糊理论应用于图像增强的方法，基于此提出了最小模糊偏移度准则下SAR图像自动对比度增强算法。利用
[Abstract]:Because of its unique coherent imaging mechanism, the imaging results of synthetic Aperture Radar (SAR) have many quality problems different from optical images, such as sidelobe noise, speckle and so on, which make the image dynamic range small, contrast low and blur. SAR image interpretation is seriously affected. The process of quality evaluation is to evaluate the quality of SAR images with objective parameters. It is a pre-processing step before image interpretation and has important research value. The main contents of this paper are as follows: 1. This paper studies the basic method of SAR image quality evaluation. The famous radar systems, such as RADARSARN ENVISATJERS and so on, have built radar data quality monitoring stations on the ground, including resolution, sidelobe intensity, geometric positioning accuracy, etc. The quality of radar echo data is preliminarily evaluated in terms of radiative resolution. In this paper, according to the different evaluation methods, these parameters are divided into point target quality parameters and surface target quality parameters, and their definitions and calculation methods are described in detail. In order to solve the problem that point target is difficult to obtain and point target quality parameter has large calculation error, a method of evaluating point target quality parameter by using coherent correlation function of complex SAR image region is presented in this paper. The simulated and measured SAR image data show that the proposed method is effective in calculating the resolution, peak sidelobe ratio and integral sidelobe ratio of SAR images. The principle of defocus blur in SAR image and the evaluation algorithm of defocus blur degree are studied. Defocus blur is one of the important causes of SAR image blur, which is mainly caused by the frequency estimation error. Based on the imaging principle of SAR, this paper analyzes the influence of defocus blur on imaging results, and obtains the conclusion that the main lobe is widened and the sidelobe is raised, which is similar to the result of motion blur of optical image. Based on the method of motion blur parameter evaluation, an effective defocus fuzzy parameter evaluation algorithm for SAR image is proposed in this paper. The Radon transform of image truncation spectrum is used to evaluate the fuzzy direction, and the median filter with large window is used for fast denoising. An algorithm for evaluating defocus ambiguity with the mean edge width of image salient region. In this paper, a defocusing ambiguity evaluation index, UBDD, is defined, and it is proved by experiments that the index can reach a conclusion consistent with the subjective evaluation. The contrast evaluation method and enhancement algorithm of SAR image are studied. Image contrast reflects the clarity of image, and determines the ability of image information acquisition. SAR image contrast evaluation is not only an important part of image quality assessment, but also can objectively evaluate the merits and demerits of image enhancement algorithm. In this paper, two contrast evaluation parameters based on human visual characteristics, EME and TEN, are described, and the effectiveness of using EME-TEN parameter combination to evaluate the contrast of SAR images is verified by experiments. The method of applying fuzzy theory to image enhancement is also studied in this paper. Based on this, an automatic contrast enhancement algorithm for SAR images based on minimum fuzzy offset criterion is proposed. Utilization
【学位授予单位】：国防科学技术大学
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：TN957.52

【参考文献】