基于无人机图像颜色指数的植被识别  

Vegetation extraction method based on color indices from UAV images

作　　者：丁雷龙 李强子[2] 杜鑫[2] 田亦陈[2] 袁超[2] 

机构地区：[1]中国地质大学(北京)地球科学与资源学院,北京100083 [2]中国科学院遥感与数字地球研究所,北京100101

出　　处：《国土资源遥感》2016年第1期78-86,共9页Remote Sensing for Land & Resources

摘　　要：植被信息在农业监测、生态环境保护等方面具有重要作用。利用无人机(unmanned aerial vehicle,UAV)获取的高分辨率图像识别植被信息具有成本低廉、方式灵活等优势。目前UAV遥感使用的可见光图像主要依靠各种颜色指数提取植被。以山东省微山县为研究区,选用NGRDI,Ex G,Ex G-ExR和GLI等4种基于RGB色域的颜色指数,对覆盖研究区的UAV图像进行灰度化处理,用最大类间方差自动阈值检测方法将植被区域与非植被区域识别出来,并分析各种颜色指数的适用性及影响因素。研究结果表明:4种颜色指数均能快速准确地识别植被覆盖区域,识别精度在90%以上。其中Ex G-ExR指数优于其他指数,识别精度最高,识别效果较稳定;Ex G与GLI指数的识别精度在研究区9景图像中变化不大,相对稳定,也可作为有效的植被识别方法。4种颜色指数对植被与背景的RGB特征差别较大图像的植被识别精度均较高。植被识别精度与研究区图像中冬小麦所占比例成正比,与阔叶林、建筑物/道路所占比例成反比。Vegetation extraction plays a key role in such aspects as agricultural monitoring, ecological and environmental function evaluation. Traditionally, it is a labor input task before remotely sensed images are involved. Yet it is difficult to deal with conditions when clouds exist. Therefore,scenes become hopeful when unmanned aerial vehicle（ UAV） images emerge in vegetation extraction,and hence this means is a low- cost and flexible way with high spatial resolution. High efficiency methods are therefore required to extract vegetation area automatically using UAV images,preferentially with various color indices always involved. The problem is that there is no evaluation of the effects of vegetation extraction with different color indices. In this paper,Weishan County of Shandong Province was chosen as the study area and four color vegetation indices comprising normalized green- red difference（ NGRDI）,excess green（ Ex G）,excess green minus excess red（ Ex G- ExR） and green leaf index（ GLI） were selected to extract vegetation information from UAV images with OTSU threshold values. The results show that all the color indices are capable of extracting vegetation area with accuracies above 90%. Ex G-ExR index could more likely generate higher accurate results than other indices. Ex G and GLI indices generate relatively high accurate and stable results,and could also be used for effective vegetation extraction. For images with high RGB value contrast between vegetation and background,all the color indices work especially well.Further analysis has revealed that accuracies of vegetation extraction have positive relationship with the proportion of winter wheat in images,and exhibit negative relationship with the proportion of broadleaf trees,buildings and roads.

关 键 词：无人机(UAV) 植被识别 颜色指数 RGB色域 

分 类 号：TP751.1[自动化与计算机技术—检测技术与自动化装置]