针对传统人工城市绿地调查费时、费力等问题,该文提出了一种利用无人机遥感进行城市绿地信息提取的思路和方法。设计构建了基于固定翼无人机平台的遥感系统,对宁波市部分区域进行航飞实验。采用PixelGrid软件对无人机影像进行了预处理,得到测区正射影像。采用基于像元统计特征分类和面向对象分类两种方法,对研究区进行绿地信息快速提取并进行了精度验证。研究结果表明,利用最大似然分类法和归一化绿红差异指数法提取精度最高,分别为81.73%和80.23%。
In order to avoid the time-consuming and laborious work of traditional manual survey of ur- ban green space, this paper put forward a new idea and surveying method. With a remote sensing system based on the fixed wing UAV, it carried on flight experiment in some areas of Ningbo city; then encrypted aerial triangulation and produced images by using PixelGrid. In the end, it extracted the green space information fast and conducted accuracy verification in the surveyed areas, with the methods of statistical feature classification and object-oriented classification. The results showed that the highest extraction accuracy was produced by using the maximum likelihood classification and normalized green-red difference index, with accuracy percentages of 81.73% and 80.23% respectively.