基于遥感影像的地块边界半自动提取  

Semi-Automatic Boundary Extraction Based on Gaussian Model

作　　者：宋建涛 李大军[1] 郭丙轩[2] SONG Jiantao;LI Dajun;GUO Bingxuan(East China University of Technology,Nanchang Jiangxi 330013,China;State Key Laboratory of Information Engineering in Surveying,Mapping and Remote Sensing,Wuhan University,Wuhan Hubei 430079,China)

机构地区：[1]东华理工大学测绘工程学院,江西南昌330013 [2]武汉大学测绘遥感信息工程国家重点实验室,湖北武汉430079

出　　处：《北京测绘》2019年第10期1171-1175,共5页Beijing Surveying and Mapping

基　　金：国家重点研发计划(2016YFB0502200)

摘　　要：针对遥感影像图中地块的提取,提出了一种基于高斯模型的地块边界的半自动提取方法。以无人机影像为实验数据,首先利用影像分割得到包含要提取地块的窗口影像;然后利用高斯模型对地块进行提取,该算法解决了传统图割算法抗噪能力差的问题。随后,运用数学形态学对分割的结果做进一步处理;最后利用边界提取函数对地块边界进行提取并对边界进行拟合。实验结果证明,本文算法不仅提取地块精度较高,而且适当的减轻了工作人员的工作强度。Aiming at the extraction of land parcel in remote sensing image,a semi-automatic extraction method of land parcel boundary based on gaussian model is proposed.Taking the uav image as experimental data,the window image containing the plot to be extracted was firstly obtained by image segmentation.Then gaussian model is used to extract the plots.This algorithm solves the problem of poor anti-noise ability of traditional graph cutting algorithm.Then,the results of segmentation are further processed by mathematical morphology.At last,boundary extraction function is used to extract and fit the boundary.The experimental results show that the proposed algorithm can not only extract the plots with high precision,but also reduce the working intensity of the workers.

关 键 词：高斯模型 数学形态学 边界提取 最小多边形拟合 

分 类 号：P237[天文地球—摄影测量与遥感]