自然资源遥感2017，Vol.29Issue(3)：77-84,8.DOI:10.6046/gtzyyg.2017.03.11
基于面向对象变化向量分析法的遥感影像森林变化检测
Forest change detection using remote sensing image based on object-oriented change vector analysis
摘要
Abstract
To develop a method for collecting spatial information of forest change to update forest resources database, the authors tested a forest change detection in an area in Shangsi County of Guangxi where the forest cover changed frequently and rapidly and had a lot of change parcels most of which were small patches.ZY-3 and GF-1 satellite remote sensing images and the thematic map of forest distribution composed of sub-compartments were used as the data sources, the length of change vector was measured by Mahalanobis distance, Euclidean distance and relative error distance, and the optimal threshold was determined by the objective function.In addition, the object-based change vector analysis (CVA)was used to detect the forest change based on the sub-compartment.The results show that the detection results based on the Mahalanobis distance and Euclidean distance are not ideal, for they have high omission rate and commission rate but low total accuracy and small kappa coefficient.The detection result based on the relative error distance is the best among the three detections, for its omission accuracy (21.0%) and the commission accuracy (32.5%) are the lowest in the three detection, and its total accuracy (89.6%) and its Kappa coefficient (0.664) are higher than the two other detections.False detections are usually found in the old forest land, construction area, road and some other places, and the commission objects are found in various land types.关键词
面向对象/变化向量分析(CVA)/目标函数/变化检测/小班Key words
object-oriented/change vector analysis(CVA)/object function/change detection/sub-compartmenet分类
信息技术与安全科学引用本文复制引用
李春干,梁文海..基于面向对象变化向量分析法的遥感影像森林变化检测[J].自然资源遥感,2017,29(3):77-84,8.基金项目
广西林业科学研究项目"森林变化遥感信息自动检测与提取"(编号: 201423)资助. （编号: 201423）
