SAR图像线特征分析与自动提取  

Analysis and automatic extraction of linear features in synthetic aperture radar images

作　　者：陈思[1,2] 杨健[1] 宋小全[2] 

机构地区：[1]清华大学电子工程系,北京100084 [2]北京跟踪与通信技术研究所,北京100094

出　　处：《系统工程与电子技术》2010年第9期1868-1874,共7页Systems Engineering and Electronics

基　　金：国家自然科学基金(40871157)资助课题

摘　　要：自动提取图像中的线特征,特别是长而直的线特征,对于实现场景的自动解译有重要意义。对合成孔径雷达(syhthetic aperture radar,SAR)图像中的线特征及其自动提取问题的特殊性进行分析,结合对已有方法共性的简单总结,从实际应用需要的角度归纳了SAR图像线特征自动提取方法应满足的基本要求和应遵循的合理思路。提出了一种由粗到细的自动提取方法,定义一组局部方向参数描述图像的局部纹理方向特征,通过多尺度分析实现快速粗提取,在后处理中进行细化并获得知识表达。在日本PI-SAR实际数据图像上进行的实验表明了提出方法的有效性。Automatic extraction of linear features in images,especially the long and straight ones,means a lot to automatic scene interpreting.The specialities of linear features and their automatic extraction in synthetic aperture radar（SAR） images are analyzed,combined with a simple summarization of the commonness of existing methods,the basic requirements and reasonable thoughts that the method of automatic extraction of linear features in SAR images should follow are summed up from the practical application needs＇ point of view.A from-coarse-to-fine method of the automatic extraction is proposed,a set of local direction parameters are defined to describe the local texture direction features of the image,the fast coarse extraction is realized by multi-scale analyzing,the refinement and the knowledge expression are achieved in the post processing.The effectivity of the proposed method is validated by experiments on the real data image of Japanese PI-SAR.

关 键 词：合成孔径雷达 线特征 局部方向参数 多尺度 

分 类 号：TN959.3[电子电信—信号与信息处理]