一种基于边缘特征的海岸线检测方法  

A Method for Coastline Extraction Based on edges

作　　者：荆浩[1] 陈学佺[1] 顾志伟[1] 

机构地区：[1]中国科学技术大学电子工程与信息科学系,安徽合肥230027

出　　处：《计算机仿真》2006年第8期89-93,101,共6页Computer Simulation

摘　　要：针对遥感图像中灰度特征的不稳定性,尤其是SAR(合成孔径雷达)图像中水陆灰度的弱对比性,提出一种利用相对稳定的边缘特征进行海岸线检测的新方法。该方法首先用Roberts算子提取海岸区域图像的梯度,并进行自动阈值分割,进而用轮廓跟踪的方法得到海岸线的粗略位置。借鉴主动轮廓的思想,设计一种轮廓逼近的方法,根据遥感图像的梯度强度对粗略的海岸线进行调整,最终得到精确的海岸线。实验表明这种方法在检测精度和抑制海域内干扰等方面都能达到令人满意的效果。In consideration of the instability of gray characteristic in remote sensing images, especially the low contrast of land - sea gray level in SAR＇s （ Synthetic Aperture Radars） images, a novel algorithm utilizing the comparatively steadier edge characteristic for coastline detection is presented. The Roberts operator is used to generate a grad map of a given coastal image, and a automatically selected threshold is applied to the grad map, then the crude position of the coastline can be obtained with a contour tracing algorithm. Referring to the idea of active contour, a method for contour attraction is designed to adjust the obtained crude coastline according to the edge strength of the original image, and accurate coastline is extracted finally. Experiments show that the algorithm performs satisfactorily in aspects such as detection accuracy and inhibition of sea interference, etc.

关 键 词：海岸线检测 合成孔径雷达 边缘特征 主动轮廓 干扰抑制 

分 类 号：TP75[自动化与计算机技术—检测技术与自动化装置]