摘要
首次将局部统计活动轮廓模型引入SAR图像海岸线检测问题中,提出了一种基于局部统计活动轮廓模型的SAR图像海岸线检测方法。首先利用C-V模型进行粗分割,消除局部统计活动轮廓模型对初始轮廓线设置要求严格的限制,然后提出了一种基于G0分布的局部统计活动轮廓模型,进行精细分割。该模型采用G0分布对轮廓线上每一点的邻域进行统计建模,增强了模型数据拟合能力,提高了海岸线检测精度,加入水平集函数惩罚项,消除了重新初始化过程。实测SAR图像实验表明,本文方法可用于精确海岸线检测。
A coastline detection method using synthetic aperture radar (SAR)images based on local statistical active contour model has been proposed in this paper. The method incorporates the local statistical active contour model to detect the coastline in SAR images. In order to remove the limitation of a rigid initial contour being requested in the local statistical active contour model, this method firstly utilizes a C-V model to gain an approximate segmentation. Thereafter, a local statistical active contour model based on G0 distribution is proposed to achieve the accurate segmentation results. The new model adopted G0 distribution to fit each neighborhood along the contour, enhancing the fitting ability for SAR images and improving the detection accuracy of the coastline. Through combining a penalizing term of level set function, the model eliminates the need of re-initialization procedure. The experiments of real SAR images demonstrate the proposed method has accurate coastline detection ability.
出处
《遥感学报》
EI
CSCD
北大核心
2011年第4期737-749,共13页
NATIONAL REMOTE SENSING BULLETIN
基金
国防科学技术大学优秀研究生创新基金资助(编号:S080501)~~