基于像元信息分解的城市绿地信息自动提取系统

 为了快速而有效地监测城市绿地的变化，建立了基于遥感和GIS的城市绿地自动识别系统。系统利用像元信息分解的方法，将城市环境下的像元信息分解为绿地、水体、土壤、水泥地等4种地物端元组份信息，求解绿地在混合像元中的丰度，并允许自由设定判定绿地像元的阈值，以便通过调整达到分离的最优分离结果。此外,系统也集成了绿地信息的空间分析以及统计分析识别等功能。最后，以深圳市罗湖区和福田区TM影像作为实验区，将系统自动提取的绿地面积和目视解译的结果进行对比，准确率达到98.7%。

An Automatic Extraction System of Urban Greenbelt based on Pixel Information Decomposition Model

New automatic classification system,using pixel information decomposition model as the key function,is constructed in order to effectively pick up the data of urban greenbelt change from remote image.Firstly,image pixel information is divided into four elementary components—water,vegetation, soil and cement ground. Then, the abundances of four components in pixel are calculated and the components are automatically decomposited by the system. In this system, the threshold value of vegetation discriminant is adjustable, so the optimal result can be obtained through adjusting the threshold value. Besides the pixel information decomposition model, the function of space analysis model and the statistic analysis model are integrated in this system. Finally, utilizing the TM image of Futian and Luohu Districts of Shenzhen city as case study, a precision of 987% is obtained when compared the greenbelt acreage extracted by pixel information decomposition model with the result obtained by judging image with eyes.