SAR图像非局部均值去噪和分割应用研究

发布时间：2018-01-01 21:06

本文关键词：SAR图像非局部均值去噪和分割应用研究　出处：《西安电子科技大学》2015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)相比于光学传感器具有多波段、视角可变、穿透力强,能全天时、全天候对地扫描等优点,在国民经济和军事领域都有着广泛的应用。然而由SAR系统成像机制所致,相干斑噪声一直是严重影响SAR图像成像质量的重要因素,给SAR图像后续的分类、分割、目标检测和识别等应用带来极大的困难。极化SAR图像属于SAR图像的一种,也饱受相干斑噪声的影响,其相干斑抑制工作一直是国际学者们研究的热点之一。图像分割也是图像处理的主要研究方向之一,机场作为重要的军事目标,SAR图像中的机场检测分割也有重要的研究价值。现有的分割方法都是依据一定的图像特征、利用特定的分割准则来完成图像的分割,然而由于待分割样本差异较大,不存在一种“最好”的算法适用于所有的分割问题,各种方法都存在各自的局限性,因此对各类分割方法的研究也一直没有间断过。对这两个方面,本文的主要工作如下:(1)提出了一种基于贝叶斯非局部均值的极化SAR图像质量增强算法。结合非局部均值模型,在贝叶斯理论框架下给出了的权值计算方法及理论证明,通过验证实验本文方法得到的权值贴近数据的真实情况,本文方法的滤波结果在相干斑抑制和细节保持方面有较好的平衡。(2)提出了一种基于局部阈值分割的机场SAR图像分割算法。机场都有明显的几何特征,本文从局部阈值分割的思想出发,在图像分块的过程中利用机场结构信息,使图像的分块更加合理,同时考虑到照度不均等因素造成的图像强度不一致问题,引入了照度补偿函数。通过对真实的机场SAR图像分割测验,本文提出的方法获得了较为理想的分割结果,达到了预期的要求。
[Abstract]:Compared with optical sensors, synthetic Aperture radar (SAR) has multi-band, variable angle of view, strong penetration, and can be used all day. All-weather ground scanning has been widely used in national economy and military field. However, it is caused by the imaging mechanism of SAR system. Speckle noise has always been an important factor affecting the imaging quality of SAR images. It classifies and segments SAR images. The application of target detection and recognition brings great difficulties. Polarimetric SAR image is a kind of SAR image, which is also affected by speckle noise. Image segmentation is one of the main research directions of image processing, and the airport is an important military target. Airport detection and segmentation in SAR images also have important research value. Existing segmentation methods are based on certain image features, using specific segmentation criteria to complete image segmentation. However, due to the large differences of samples to be segmented, there is no "best" algorithm suitable for all segmentation problems, and each method has its own limitations. Therefore, the research on all kinds of segmentation methods has not been interrupted. The main work of this paper is as follows: (1) A new image quality enhancement algorithm based on Bayesian non-local mean is proposed, which combines the non-local mean model. Under the framework of Bayesian theory, the weight calculation method and the theoretical proof are given, and the experimental results show that the weights obtained by this method are close to the real situation of the data. The filtering results of this method have a good balance in speckle suppression and detail preservation.) A segmentation algorithm for airport SAR image based on local threshold segmentation is proposed. In this paper, based on the idea of local threshold segmentation, the field structure information is used in the process of image segmentation, which makes the image segmentation more reasonable, and takes into account the inconsistency of image intensity caused by the inhomogeneity of illumination. The illumination compensation function is introduced. By testing the real airport SAR image segmentation, the method proposed in this paper obtains more ideal segmentation results and meets the expected requirements.
【学位授予单位】：西安电子科技大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TN957.52

【参考文献】