基于GF-2影像的面向对象植被信息提取研究

Research on Object Oriented Vegetation Information Extraction Based on GF-2 Image

基于GF-2遥感影像,通过面向对象方法对长株潭城市群生态绿心区内具有代表性的区域进行植被信息提取,利用专业分割软件对遥感影像进行多尺度分割,采用ESP工具定量获取多尺度分割参数,最后通过隶属度函数对遥感影像分类进行植被信息提取,并进行精度评价。结果表明:采用面向对象的方法,从高空间分辨率遥感影像中挖掘出能提取植被的光谱、纹理、形状等特征,能实现植被信息的快速提取,提取总体精度达85.33%。

Based on the GF-2 remote sensing image,vegetation information is extracted from the representative areas of the ecological green center of the Changsha-Zhuzhou-Xiangtan city group by the object-oriented method,multi-scale segmentation to remote sensing image is carried out by using the professional segmentation software,and multi-scale segmentation parameters are obtained quantitatively by using the ESP tool.Finally,vegetation information is extracted from the remote sensing image classification by the membership function and the accuracy is evaluated.The results show that the object-oriented method can extract the spectrum,texture,shape and other features of vegetation from the high spatial resolution remote sensing image,and achieve the fast extraction of vegetation information,with the overall accuracy of 85.33%.