基于活动轮廓模型的SAR图像海岸线检测  

Shoreline Detection by SAR Images Based on Active Contour Model

作　　者：邓竣天 王小龙 DENG Juntian;WANG Xiaolong

机构地区：[1]中国科学院空天信息创新研究院,北京100094 [2]中国科学院大学电子电气与通信工程学院,北京100094

出　　处：《科技创新与应用》2024年第28期1-7,共7页Technology Innovation and Application

摘　　要：SAR图像岸线检测是SAR近岸海洋目标检测的一项重要环节。该文提出一种SAR图像海岸线检测方式,该方法基于自适应混合活动轮廓模型,旨在对灰度不均匀SAR图像的海岸线进行检测。首先对SAR图像进行显著性检测的预处理,然后将预处理后的结果代入混合活动轮廓模型中进行检测。该文采用改良后的CV模型和LIF模型并以其各自的自适应参数为基础进行融合,从而增强整个模型的识别性能。从SAR图象试验的结果来看,这种方式相较于传统的CV模型技术,对灰度分布不均一的SAR图像具有更强的适用性和计算速度上的优化。Using SAR images in shoreline detection is an important part of SAR coastal ocean target detection.A shoreline detection method for SAR images is proposed.The method is based on an adaptive hybrid active contour model and aims to detect shoreline in SAR images with uneven gray levels.First,SAR images are preprocessed for saliency detection,and then the preprocessed results are substituted into the mixed active contour model for detection.In this paper,the improved CV model and LIF model are used and fused based on their respective adaptive parameters,thereby enhancing the recognition performance of the entire model.Judging from the results of SAR image experiments,compared with the traditional CV model technology,this method has stronger applicability and optimization in computing speed for SAR images with uneven gray distribution.

关 键 词：合成孔径雷达(SAR) CV模型 显著性检测 混合活动轮廓模型 识别性能 

分 类 号：TN957.52[电子电信—信号与信息处理]