一种面向对象下城市可见光航空影像植被信息提取方法

Object-oriented method for extracting vegetation information from urban visible light aerial image

无人机航空影像一般仅包含红、绿、蓝3种波段,已成为快速小范围信息提取的主要数据源。可见光植被指数是可见光航空影像进行植被信息提取的主要方法,针对可见光植被指数在复杂城市区域进行植被信息提取时存在植被与水体、植被与阴影不易区分的问题,本文提出了一种面向对象下结合可见光植被指数、亮度信息、纹理方差信息建立知识规则的方法进行植被信息提取,并与常用可见光植被指数结果对比,结果表明,本文方法在植被与水体和阴影部分的区分较好,植被提取的精度更高,可作为城市航空影像植被提取的一个有效方法。

Aerial images of unmanned aerial vehicles(UAVs)generally only contain three bands:red,green and blue,which has become the main data source for rapid information extraction in a small range.Visible vegetation index(VDI)is the main method for vegetation information extraction from visible aerial images.Aiming at the problem that it is difficult to distinguish vegetation from water body,vegetation and shadow when VDI is used for vegetation information extraction in complex urban areas,This paper puts forward an object-oriented under combined with visible light vegetation index,brightness,texture variance information to establish knowledge rules of vegetation information extraction method,and compared with commonly used vegetation index of visible light.The results show that the method in the difference of vegetation and water body and the shaded part is good,the distinction between vegetation extraction accuracy is higher,Which can be used as an effective method to extract vegetation from urban aerial image.