SAR图像海岸线检测的区域距离正则化几何主动轮廓模型

SAR Image Coastline Detection Based on Regional Distance Regularized Geometric Active Contour Models

合成孔径雷达(SAR)卫星遥感图像可以极大地提高全国海岸线覆盖频率,然而受到海洋波浪所引起的随机海水表面粗糙度的影响,海岸目标与海水背景边界易混淆不清,因此本文提出了基于区域距离正则化几何主动轮廓模型(RDRGAC),引入距离正则项,解决重复初始化水平集函数为符号距离函数的问题,提高了算法收敛速度。此外,将区域面积项系数与SAR图像等效视数(ENL)建立非线性拟合关系,实现RDRGAC模型根据不同SAR遥感图像的自适应调整,改善海岸线自动提取精度。通过河北省北戴河和大连市金州湾SAR数据海岸线提取对比试验,验证了所提方法的有效性。

Synthetic aperture radar（ SAR） satellite remote sensing images can greatly increase the frequency of the coastline coverage all over the country. However,due to the influence of the random sea surface roughness caused by waves,it＇s a challenge to distinguish the coastline and sea boundary. To solve this problem,this paper proposes regional distance regularized geometric active contour models（ RDRGAC）,in which the distance regularized term is introduced to avoid periodically initializing the degraded function with a signed distance function,accelerating the speed of convergence. Besides,it establishes the nonlinear regression relationship between the regional term parameters and equivalent number of looks（ ENL） in SAR images,leading to the adaptive setting of RDRGAC model with different SAR images,which could improve the accuracy of coastline automatic detection in return. In the experiments,SAR images in Beidaihe and Dalian Jinzhou Bay respectively are adopted to detect the coastline,verifying the effective of the proposed method.