基于无人机可见光遥感影像的耕地精准分类方法研究

无人机可见光遥感具有使用成本低、操作简单、实时获取遥感影像、地面分辨率高等优势。提出了一种利用无人机可见光遥感影像进行耕地精准分类的方法,以广东省惠州市惠东县铁涌镇石桥村部分耕地的可见光遥感影像为研究对象,对耕地的面积信息、形状信息以及位置信息进行监测和提取,采用面向对象法对影像中两种基于可见光波段的植被指数、纹理信息、形状信息进行分析,研究出分类提取耕地信息的较佳方案。经过反复实验确定分割尺度45、合并尺度90为分割参数,同时利用波段信息和纹理信息对未种植作物耕地和其他地物进行分离。该方法总体精度为89.23%,Kappa系数为0.72。实验结果表明利用无人机可见光遥感数据对耕地进行分类虽然存在一些细碎地块被错提、误提的情况,但总体精度仍然保持在一个很高的水准,可以为耕地作物分类提供参考,为实现精准农业提供精准的数据基础。

Precise Classification of Cultivated Land Based on Visible Remote Sensing Image of UAV

Visible remote sensing of UAV has advantages of low cost, simple operation, real-time acquisition of remote sensing images and high ground resolution. This paper proposed a method of precise classification of cultivated land based on UAV visible remote sensing image. Taking the visible remote sensing image of some cultivated land in Shiqiao Village, Tieyong Town, Huidong County, Guangdong Province as the research object, the area information, shape information and location information of cultivated land were monitored and extracted. Two kinds of vegetation based on visible wave band were used in the image by object-oriented method. Based on the analysis of index, texture and shape information, the better scheme of extracting cultivated land information by classification was studied. After repeated experiments, the segmentation scale 45 and the merging scale 90 were determined as the segmentation parameters. At the same time, the non-planted cropland and other land objects were separated using band information and texture information. The overall accuracy of the method was 89.23%, and the Kappa coefficient was 0.72. The experimental results showed that although there were some cases of mistaken lifting and mistaken lifting of fine plots in the classification of cultivated land using UAV visible remote sensing data, the overall accuracy remained at a high level, which provided a reference for crop classification of cultivated land, and provided precise data basis for precision agriculture.