QQ客服 官方微博 反馈留言

	The methodology of detailed vegetation classification based on environmental knowledge and remote sensing images
	Xu Z. G. ; Zhuang D. F. ; Ieee
	2007
关键词	detailed vegetation classification remote sensing image environmental knowledge object parcel decision free cover
页码	2074-2077
英文摘要	Detailed vegetation types information is required for many ecology and hydrology models. We can obtain the information using remote sensing image classification. However, the accuracy of the traditional classification is limited and unsatisfied. Knowledge of the relationship between vegetation and environmental factors can be utilized to assist remote sensing classification. The paper proposes a method combining this kind of knowledge and feature extracted from remote sensing images to classify vegetation types. The method consists of three steps. The first is to extract feature such as spectrum information, texture, and geometry shape from fusion of Aerial and TM images. In order to extract the information more precisely, we combined the high spatial resolution aerial image with the multi-spectral TM of satellite remote sensing image, and employed a remote sensing image fusion method based on wavelet packet transform. The second step was to derive environmental knowledge which control vegetation growth. Soil types, temperature, precipitation and landform were chosen as the main environmental factors controlling vegetation growth. Landform indexes such as elevation, slope and aspect were derived from high resolution DEM. The third step was to classify vegetation types using decision tree method with environmental knowledge and feature extracted from remote sensing images. In our research, we calculated the eigenvalue of every feature through training sample. The features included spectrum information, texture, geometry shape, soil types, temperature, precipitation, elevation, slope and wetness index. Adopting the threshold method to delaminate and extract the vegetation information according to definite threshold range, knowledge rules and decision tree. After that we classified vegetation information. A case study was done in Beijing suburb area. We used field point data to validate our classification maps. The point sampling strategy was used to obtain field vegetation data for the assessment of the overall accuracy. For point sampling, a stratification technique was employed. The final accuracy is 85%. By comparison with maximum likelihood classification the accuracy was increased by about 10%. This study showed that our method has a high accuracy and high spectral and spatial resolutions. It was able to provide more accurate vegetation classification information than traditional method.
收录类别	CPCI
会议录出版者	Ieee
语种	英语
ISBN号	978-1-4244-1211-2
内容类型	会议论文
源URL	[http://ir.igsnrr.ac.cn/handle/311030/25347]
专题	地理科学与资源研究所_历年回溯文献
推荐引用方式 GB/T 7714	Xu Z. G.,Zhuang D. F.,Ieee. The methodology of detailed vegetation classification based on environmental knowledge and remote sensing images[C]. 见:.

个性服务

查看访问统计

相关权益政策

暂无数据

收藏/分享

除非特别说明，本系统中所有内容都受版权保护，并保留所有权利。