摘要
为探索哨兵2号影像提取土地利用的可行性,研究基于像元与面向对象分类方法协同的土地资源监测技术,并选取柬埔寨上丁省为例开展应用。通过非监督分类方法将研究区划分为水体、植被和其他三大类;以其他地类为基底,利用半自动化技术提取交通运输用地信息;进一步利用多尺度分割方法完成不同区域影像切割形成细小地块对象;选取少量样本,基于传统特征指数、红边特征指数、亮度、波段均值等特征,利用面向地块对象的支持向量机迭代分类方法完成土地细类识别。结果表明,总体精度为90.24%,Kappa系数为0.89,能够满足使用要求。统计显示,林地面积最多,占领土面积的77.70%,耕地次之,占比为14.12%。实验说明哨兵2号影像用于监测土地资源切实可行,对于提高分类精度具有促进作用。文章研究方法对于实现柬埔寨土地资源监测具有参考价值。
In order to explore the feasibility of extracting land use from Sentinel-2 image,the land resource monitoring technology based on pixel and object-oriented classification method is studied,and the application of the province of Shangding,Cambodia is selected.By using the unsupervised classification method,the study area is divided into three main categories:water body,vegetation and others,using other land types as the basis,using semi-automatic technology to extract transportation land information.Furthermore,the multi-scale segmentation method is used to cut the images of different regions to form small parcel objects.After selecting a small number of samples,based on the characteristics of traditional feature index,red edge feature index,brightness,band mean,etc.,the support vector machine iterative classification method for the block object is used to complete the land classification.The results show that the overall accuracy is 90.24%,and the Kappa coefficient is 0.89,which can meet the requirements of use.Statistics show that the forest area is the largest,accounting for 77.70%of the territory,followed by cultivated land,accounting for 14.12%.The experiment shows that the sentinel-2 image is really feasible for monitoring land resources,and it can promote the classification accuracy.The research method of this paper has reference value for realizing the monitoring of land resources in Cambodia.
作者
谢国雪
穆琳
苏秋群
张家玫
苏秀刚
XIE Guoxue;MU Li;SU Qiuqun;ZHANG Jiamei;SU Xiugang(Agricultural Science and Technology Information Research Institute,Guangxi Academy of Agricultural Sciences,Nanning 530007,China;Business School,Guangxi University,Nanning 530000,China)
出处
《测绘与空间地理信息》
2020年第12期46-50,54,共6页
Geomatics & Spatial Information Technology
基金
广西创新驱动发展专项资金项目——柑橘黄龙病综合防控技术研究与示范(桂科AA18118046)
广西创新驱动发展专项资金项目——糖业发展大数据平台开发(桂科AA18118048)
国家社科基金青年项目——基于大数据及遥感的“一带一路”新闻热点及争议事件研究(19CXW028)资助。
关键词
遥感
哨兵2号
土地资源
监测技术
上丁省
remote sensing
Sentinel-2
land resources
monitoring technology
Shangding
