摘要
鉴于高分辨率遥感影像的光谱混淆和噪声干扰现象为遥感信息的精确提取带来很大困难,而利用现有矢量数据作为约束条件是提高遥感信息提取准确性的有效手段,提出了一种基于矢量约束实现面向对象高分辨率遥感影像水体提取的新方法。方法的基本步骤包括:1)利用硬边界约束方法(HBC-SEG)进行图像分割;2)分割图层与矢量数据叠合以建立分割图斑和已有矢量水体图斑间的空间关系;3)针对矢量与影像间的时相差异和配准误差建立差异化的知识规则以提取影像中的真实水体。实验证明,该方法在准确提取高分辨率影像上细小水体信息的同时,也能有效抑制建筑等地物的阴影干扰,和无约束方法相比,综合精度评价指标F-Measure平均提高13%以上。
The spectral confusion and image noises bring significant difficulties for information extraction from high spatial resolution(HSR)remote sensing images.Thus,the utilization of vector data as a constraint is feasible to improve the accuracy of remote sensing information extraction methods.This paper proposes a new object-based water-extraction method by involving vector data constraints.The basic steps of the method include:1)image segmentation is conducted using the hard boundary constraint segmentation method(HBC-SEG);2)segmentation layer and vector data are overlapped,and then topology relationships between image segments and water polygons are built up;3)differentiated knowledge rules considering the time difference and the registration errors between vector and the image data are established to extract water bodies from HSR images.Experimental results show that the proposed method can accurately extract small water bodies and suppress the interference of building shadows;the comprehensive evaluation index F-measure increases by more than 13%compared with the counterpart with no constraint.
作者
崔齐
王杰
汪闽
张东
孙艳霞
CUI Qi;WANG Jie;WANG Min;ZHANG Dong;SUN Yanxia(Key Laboratory of Virtual Geographic Environment,Ministry of Education,Nanjing Normal University,Nanjing 210023,China)
出处
《遥感信息》
CSCD
北大核心
2018年第4期115-121,共7页
Remote Sensing Information
基金
国家重点研发计划(2017YFB0503902)
国家自然科学基金(41671341)
关键词
水体
高分辨率遥感
矢量约束
遥感信息提取
面向对象图像分析
water body
high resolution remote sensing
vector data constraint
remote sensing information extraction
object-based image analysis
作者简介
崔齐(1994—),男,硕士研究生,主要研究方向为遥感数字图像处理。E-mail:cuiqi0108@163.com;通信作者:汪闽(1975—),男,博士生导师,主要研究方向为遥感信息分析、空间数据挖掘。E-mail:sysj0918@126.com
