摘要
论文提出了面向地类图斑对象提取大宗农作物信息的方法。以广西隆安县为例,提取了甘蔗、水稻、木薯、香蕉4种大宗农作物信息,分类总精度达96.2783%,Kappa系数达0.9471,能够满足使用要求。实验结果表明引入地类图斑数据能够有利于提高广西丘陵地区大宗农作物信息提取的精度。
The method to extract large crops information oriented to the land type pattern spot has been proposed in this paper. Taking Guangxi Longan County as an example, the paper extracts from sugarcane, rice, cassava, and banana 4 staple crops information. Classification accuracy is 96.2783%, and the Kappa coefficient is 0.9471.It can meet the use requirements. Experiments show that the introduction of the land type pattern spot data can improve Guangxi hilly region staple crops the accuracy of information extraction.
出处
《国土资源信息化》
2015年第3期35-41,共7页
Land and Resources Informatization
关键词
面向对象
地类图斑
影像特征
SVM模型
提取
大宗农作物
Object oriented
Class diagram spot
Image feature
the SVM model
Extraction
Staple crops