基于重采样的SAR图像边缘检测的研究

发布时间：2018-01-10 17:17

本文关键词：基于重采样的SAR图像边缘检测的研究　出处：《天津理工大学》2017年硕士论文　论文类型：学位论文

【摘要】：合成孔径雷达(Synthetic Aperture Radar,SAR)是一种主动式微波传感器,由于其具有全天时、全天候、高分辨率的特点,已被广泛应用于经济、军事和科学研究等多领域。但是由于SAR图像中固有的斑点噪声影响了图像的质量,导致SAR图像的理解与解译能力相对滞后。为了可以准确的表达SAR图像所包含信息,本文综述了SAR图像统计建模研究中的概率分布模型,并根据G~0分布的统计性质进一步研究了该模型的参数估计,最后在此基础上进行了边缘检测和置信区间估计的研究,主要内容如下:1)针对SAR图像在G~0统计模型下的参数估计,比较分析了传统的最大似然参数估计和矩估计等算法。为解决G~0分布参数估计的准确性和有效性,本文利用随机加权法的优点,提出了基于随机加权与类比估计相结合的参数估计方法,并通过不同视数和粗糙参数值在真实数据和模拟数据下进行实验验证分析,实验结果验证了本文提出的估计方法精准高、鲁棒性好。2)在进行G~0分布模型的参数估计下,本文将SAR图像边缘检测和边缘置信区间估计的两种方法进行了比较,一是基于广义似然检测和随机加权的SAR图像边缘检测和边缘置信区间估计的参数方法,二是基于KW方法和Bootstrap的SAR图像边缘检测和边缘置信区间估计的非参数方法,并通过实验验证了该方法的有效性和准确性。
[Abstract]:Synthetic Aperture radar (SAR) is an active microwave sensor, because of its all-day, all-weather, high-resolution characteristics. It has been widely used in many fields, such as economy, military and scientific research, but the image quality is affected by the inherent speckle noise in SAR image. In order to accurately express the information contained in SAR images, this paper summarizes the probabilistic distribution models in the statistical modeling of SAR images. The parameter estimation of the model is further studied according to the statistical properties of the Gn0 distribution. Finally, the edge detection and confidence interval estimation are studied. The main contents of this paper are as follows: (1) the parameter estimation of SAR image under Gn0 statistical model. The traditional maximum likelihood parameter estimation and moment estimation algorithms are compared and analyzed. In order to solve the accuracy and validity of parameter estimation of Gn0 distribution, the advantage of random weighting method is used in this paper. A parameter estimation method based on the combination of random weighting and analogy estimation is proposed, and the experimental verification and analysis are carried out under the real data and simulation data with different visual numbers and rough parameter values. The experimental results show that the proposed estimation method is accurate and robust. 2) under the condition of parameter estimation of Gn0 distribution model. In this paper, two methods of edge detection and edge confidence interval estimation in SAR images are compared. One is a parameter method based on generalized likelihood detection and random weighted SAR image edge detection and edge confidence interval estimation. The second is a non-parametric method for edge detection and edge confidence interval estimation of SAR images based on KW method and Bootstrap. The validity and accuracy of the method are verified by experiments.
【学位授予单位】：天津理工大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TN957.52

【相似文献】