结合相关系数和特征分析的植被区域自动变化检测研发  

Research and development of automatic detection technologies for changes in vegetation regions based on correlation coefficients and feature analysis

作　　者：潘建平[1] 徐永杰 李明明 胡勇 王春晓 PAN Jianping;XU Yongjie;LI Mingming;HU Yong;WANG Chunxiao(School of Civil Engineering,Chongqing Jiaotong University,Chongqing 400074,China;Chongqing Institute of Surveying and Monitoring for Planning and Natural Resources,Chongqing 401123,China;Hainan Basic Geographic Information Center,Ministry of Natural Resources,Haikou 570203,China)

机构地区：[1]重庆交通大学土木工程学院,重庆400074 [2]重庆市规划和自然资源调查监测院,重庆401123 [3]自然资源部海南基础地理信息中心,海口570203

出　　处：《自然资源遥感》2022年第1期67-75,共9页Remote Sensing for Natural Resources

基　　金：国家自然科学基金项目“智能手机辅助的多测站地面激光扫描森林调查方法研究”(编号:41801394);重庆市规划和自然资源局科技项目“面向城镇新增建设用地的遥感智能变化发现研究与应用”(编号:KJ-2020010);贵州省地矿局2019年局管科研项目“山水林田湖草生态环境监测关键技术研究与应用——以百里杜鹃为例”(编号:黔地矿科合201909)共同资助。

摘　　要：地表变化检测是遥感大数据应用的重要内容,为此以植被区域为研究对象,结合相关系数和特征分析提出图斑级自动变化检测应用方法并研发软件。该方法结合光谱和纹理特征构建地物相关系数,采用相似性度量方式进行植被区域的变化检测;然后通过分析植被与其他地物类型之间的光谱差异,选择红光波段比值进行伪变化去除;最后基于.NET框架和ArcGIS Engine二次开发组件库,设计研发了一款变化检测工具软件。导入实验数据进行变化检测,实验结果表明软件的变化检测达到94.3%的正确率和8.5%的漏检率,相对人工交互解译,软件自动化水平较高。研究结果表明所提方法和软件具有良好的应用价值。Surface change detection is an important component of the applications of remote sensing big data.However,it is essentially subject to manual interactive interpretation in actual production.With this regard,this paper developed an application method and software for the automatic detection of changes in vegetation regions on a polygon scale using correlation coefficients and feature analysis.The details are as follows.Correlation coefficients of surface features were constructed using spectral and textural features,and then the changes in vegetation regions were detected using the similarity measurement method.According to the analysis of spectral differences between the vegetation and other types of surface features,the red band ratio was selected to remove spurious changes.Finally,the change detection software was designed and developed using the.NET framework and the ArcGIS Engine component library for secondary development.Test data were imported into the software for change detection.The test results show the accuracy rate and omission rate of the software in the change detection were 94.3%and 8.5%,respectively.Furthermore,the software has a higher automatic level compared to manual interactive interpretation.In conclusion,the method and software developed in this study can be widely applied.

关 键 词：变化检测 相关系数 植被 图斑 特征分析 

分 类 号：TP79[自动化与计算机技术—检测技术与自动化装置]