结合相干斑抑制和区域生长的SAR海冰图像MRF分割方法研究

发布时间：2018-01-23 21:15

本文关键词： 合成孔径雷达海冰分割 SRBF 区域生长 MRF　出处：《合肥工业大学》2014年硕士论文　论文类型：学位论文

【摘要】：海冰是全球气候系统的重要组成部分,海冰信息对气候变化的监测和人类活动具有非常重要的意义。合成孔径雷达(Synthetic Aperture Radar, SAR)是一种有效的海冰监测工具,能够全天时、全天候获取大范围的高分辨率图像,在海冰研究中得到了广泛应用。基于MRF模型的分割方法是常用的SAR海冰图像分割方法。然而由于SAR海冰图像受相干斑噪声影响十分严重,使得MRF模型不能准确描述海冰图像的统计特征,从而导致分割的准确性降低。针对此问题,本文在充分研究SAR海冰图像特性的基础上,建立了相干斑抑制区域生长(SRRG)模型,将其与区域MRF相结合得到更有效的SAR海冰图像分割算法。本文研究工作主要内容如下：1、建立相干斑抑制区域生长(SRRG)模型。SRRG模型由图像的相干斑抑制区域化表达的构建和基于区域的灰度相似性的区域生长两个部分组成。相干斑抑制的区域化表达由相干斑抑制的双边滤波(SRBF)算法和分水岭变换构成。在相干斑噪声严重的情况下,SRRG模型能够有效抑制过分割,并对目标边缘准确定位。2、将SRRG模型与区域MRF结合获得有效的SAR海冰图像分割算法。SRRG模型与区域MRF的结合可以大幅度减小优化搜索空间,防止能量函数最优解陷入局部极小值,从而获得准确的分割结果。3、分别采用不同噪声程度的合成SAR海冰图像、由RADARSAT-2获取的辽东湾海冰图像和由SIR-C获取的拉布拉多海岸C波段的海冰图像进行测试,对本文给出的SRRG模型和SRRG-MRF分割算法进行充分的实验分析。结果表明,与已有的区域MRF分割算法相比,基于SRRG-MRF的SAR海冰图像分割算法能够有效提高分割准确性,证明了算法的有效性和可行性。
[Abstract]:Sea ice is an important component of the global climate system. Sea ice information is of great significance to the monitoring of climate change and human activities. Synthetic Aperture Radar synthetic Aperture Radar. Sare is an effective sea ice monitoring tool, which can obtain large range of high-resolution images all day and day. It is widely used in the research of sea ice. The segmentation method based on MRF model is commonly used in SAR sea ice image segmentation. However, the SAR sea ice image is seriously affected by speckle noise. The MRF model can not accurately describe the statistical features of sea ice image, which leads to the accuracy of segmentation reduced. In order to solve this problem, this paper fully studies the characteristics of SAR sea ice image. A speckle suppression region growth model is established and a more effective SAR sea ice image segmentation algorithm is obtained by combining it with regional MRF. The main contents of this paper are as follows: 1. Establishment of SRRG for speckle suppression region. SRRG model consists of two parts: the construction of speckle suppression regionalization expression of image and the region growth based on region gray similarity. The regionalization expression of speckle suppression is derived from the bilateral filtering of speckle suppression. Wave (. SRBF algorithm and watershed transform. In the case of serious speckle noise. The SRRG model can effectively suppress over-segmentation and locate the target edge accurately. 2. The combination of SRRG model and regional MRF to obtain an effective SAR sea ice image segmentation algorithm. SRRG model and regional MRF can greatly reduce the optimal search space. To prevent the optimal solution of energy function from falling into the local minimum, the accurate segmentation result .3is obtained, and the SAR sea ice images with different noise levels are synthesized respectively. The sea ice images of Liaodong Bay and Labrador Coast obtained by RADARSAT-2 and SIR-C were tested. The experimental results of the SRRG model and the SRRG-MRF segmentation algorithm are given in this paper. The results show that the proposed algorithm is compared with the existing region MRF segmentation algorithm. The SAR sea ice image segmentation algorithm based on SRRG-MRF can effectively improve the accuracy of the segmentation, which proves the validity and feasibility of the algorithm.
【学位授予单位】：合肥工业大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：P715.7

【相似文献】