基于IKONOS的人工林地识别研究

The Recognition of Planted Woodland Using IKONOS Data

以五分地沟试验区为例,提出了一种基于IKONOS卫星遥感图像提取不同郁闭度人工林地的方法。利用Matlab软件图像处理工具箱,进行树木信息的提取,精度达90.7%,KHAT统计量为0.86。以树木分类结果为基础,依据郁闭度定义,通过滤波方法识别不同的林地类型。结果表明,综合运用遥感数字图像处理软件、GIS软件和Matlab,能够从高分辨率的遥感数据中自动且有效地识别人工杨树林、人工杨树疏林和散生树等植被类型,可靠性较高,且可以重复验证。

According to the case study in Wufendigou, a method of discriminating planted woodland with different canopy densities was put forward by use of IKONOS digital image, a kind of remotely sensed data with high ground resolution. Firstly, trees were extracted from background by means of vegetation index derived from the image processing toolbox of Matlab software. The accuracy of tree classification was 90.7%, and k statistic was up to 0.86. And then, with the tree classification image, different types of woodland were recognized by use of filtering coupled with definitions of woodland, open woodland, and dispersed trees. The result showed the remotely sensed data with high ground resolution could be used to generate distributions of planted woodland and open woodland dominated by Populus sp. , and dispersed trees by flexible using of different kinds of software, such as digital image processing of remote sensing, GIS, and Matlab. The method proposed in this study showed a high degree of reliability and could be repeated as well. Therefore, the method had a great potential for the investigation and management of forestry re- sources.