摘要
受地理环境和人类活动影响,实时、准确地获取用户满意的地表覆盖信息一直是亟待解决的难题。针对传统单分类器对地表覆盖信息实时提取不足,提出了一种基于K-means聚类投票的地表覆盖信息快速提取方法。结合TIFF与ASCII数据格式转换,对试验区湖北省黄石市2017年的遥感影像进行地表覆盖信息提取,并对各分类器的分类精度进行了评估。结果表明,集成投票法能够快速准确地表达黄石市的地理信息,分类精度高达95.87%,Kappa系数为0.88,明显高于其他3种分类器,验证了该方法的可行性,对地表覆盖遥感制图具有一定的工程实践意义。
Affected by the geographical environment and human activities,real-time and accurate acquisition of user-satisfied surface coverage information has always been an urgent problem to be solved.Aiming at the insufficient real-time extraction of land cover information by traditional single classifiers,a fast extraction method of land cover information based on K-means cluster voting was proposed.Based on TIFF and ASCII data format conversion,the surface coverage information was extracted from the remote sensing image of Huangshi City,Hubei Province in 2017 and the classification accuracy of each classifier was evaluated.The results showed that the integrated voting method can quickly and accurately express the geographic information of Huangshi City.With the classification accuracy as high as 95.87%,the Kappa coefficient was 0.88,which was significantly higher than the other three classifiers.This experimental results verified the feasibility of the method and provide certain engineering practical significance for the remote sensing mapping of surface covering.
作者
瞿珊珊
康顺
QU Shanshan;KANG Shun(School of Electrical and Electronic Information Engineering,Hubei Polytechnic University,Huangshi Hubei 435003)
出处
《湖北理工学院学报》
2021年第4期25-28,68,共5页
Journal of Hubei Polytechnic University
基金
湖北理工学院青年项目(项目编号:20xjz06Q,320202100304)。
