结合马尔可夫随机场和混合模型的海岸线提取  

Coastline Extraction Based on SAR images CombiningMarkov Random Field and Mixture Model

作　　者：李淑瑾 石雪 钟炜 陆骏飞 LI Shujin;SHI Xue;ZHONG Wei;LU Junfei(College of Surveying and Geographic Informatics,Guilin University of Technology,Guilin,Guangxi 541004,China)

机构地区：[1]桂林理工大学测绘地理信息学院,广西桂林541004

出　　处：《遥感信息》2024年第1期167-174,共8页Remote Sensing Information

基　　金：广西自然科学基金(2022GXNSFBA035567)。

摘　　要：为了高效且准确地实现基于遥感影像的海岸线提取,提出一种结合马尔可夫随机场和混合模型的合成孔径雷达(synthetic aperture radar,SAR)影像海岸线提取算法。该算法以统计模型理论为研究基础,考虑SAR影像中同一地物像素反射强度的统计分布具有非对称和重尾的统计特性,利用伽马混合模型建立SAR影像内像素强度的概率分布。为了建模像素的空间相关性,采用马尔可夫随机场构建伽马混合模型的组分权重概率分布以克服SAR影像相干斑噪声的影响。结合马尔可夫随机场和伽马混合模型构建出SAR影像海陆分割模型,通过最大期望方法估计模型参数以实现准确的海陆分割,进而实现海岸线提取。在Sentinel-1卫星SAR影像上进行海岸线提取实验,实验结果表明该算法可实现准确的海岸线提取。To efficiently and accurately extract the coastline based on remote sensing images,a new coastline extraction algorithm combining Markov random field and mixture model is proposed in this paper.The propose algorithm is based on the statistical model theory.Due to the asymmetric and heavy-tailed characteristics of the statistical distribution of pixel reflection intensity in the same ground object of synthetic aperture radar(SAR)images,Gamma mixture model is used to build the probability distribution of pixel intensity in SAR images.Considering the spatial correlation of local pixels,Markov random field is utilized to model the distribution of the component weight of gamma mixture model to overcome the effect of speckle noise.Then,SAR image segmentation model is built by combining MRF and gamma mixture model.Finally,parameter estimation is achieved using expectation-maximization to extract the coastline.To verify the performance of the proposed algorithm,the Sentinel-1 SAR image for Beibu Gulf is selected for experiment.The experimental results show that the proposed algorithm can accurately extract the coastline.

关 键 词：海岸线提取 SAR影像分割 马尔可夫随机场 伽马混合模型 最大期望算法 

分 类 号：TP751[自动化与计算机技术—检测技术与自动化装置]