基于混合Wishart分布MRF分割的极化SAR图像海岸线检测

针对基于Wishart分布马尔可夫场(Markov random field, MRF)海陆分割存在海面和陆地整体区域无法使用单一Wishart分布描述的问题,提出了一种基于混合Wishart分布的极化合成孔径雷达(synthetic aperture radar, SAR)图像海岸线检测方法。该方法首先对极化SAR总功率边缘能量采用两区域Ostu阈值分割得到初始海陆分割结果,然后采用混合Wishart分布描述陆地和海面区域,通过基于混合Wishart分布MRF两区域分割迭代计算实现海陆精确分割。最后对经过水域合并处理的海陆分割结果进行边界跟踪实现海岸线检测。分别使用了RADARSAT2中国海南陵水地区和新加坡部分地区极化SAR数据进行实验,实验结果证明提出方法比基于Wishart分布MRF分割方法更加精确和鲁棒。

Coastline detection in polarimetric SAR images using Markov random field segmentation based on mixture Wishart distribution

The probability distribution of the sea or land is difficult to be modeled by a single Wishart distribution when segmenting the sea and land with markov random field (MRF) segmentation method in polarimetric synthetic aperture radar (SAR) imagery, a novel coastline detection method with MRF segmentation based on mixture Wishart distribution is proposed in this article. An initial sea-land segmentation result is obtained by thresholding segmentation of the edge energy map of the Span of polarimetric SAR image with the Ostu method at first. Then accurate sea-land segmentation is carried out by using the two-region MRF segmentation model, in which the distribution of the sea and land are modeled by a mixture of Wishart distribution. Finally, coastlines are extracted by tracing the boundary of the sea-land segmentation result after merging different water regions to get the final segmentation result. Polarimetric data acquired by TerraSAR over Sanfranciso region and RADARSAT2 over Singapore region are used for testing. Experiment results show that the proposed method is more accurate and robust compared with the MRF segmentation based on a single Wishart distribution.