基于无人机可见光波段对荒漠植被覆盖度提取的研究——以沙坡头地区为例

Extraction of desert vegetation coverage based on visible light band information of unmanned aerial vehicle:a case study of Shapotou region

以腾格里沙漠东南缘沙坡头地区为研究区域,采用无人机(Phantom 3A)拍摄获取指定范围可见光影像,通过ENVI软件分析照片的红、绿、蓝信息进行植被指数的提取和植被覆盖度的计算,主要分析了可见光波段差异植被指数、归一化绿化差异指数、归一化绿蓝差异指数、过绿指数、红绿比值指数与研究区荒漠植被覆盖度回归模型,探究最优反演模型,目的是验证在相同区域、相同时间拍摄的无人机影像的植被覆盖度.计算出来的植被覆盖度和无人机航片的处理方法进行比对,验证反演模型的正确性.结果表明:通过无人机可见光波段提取的植被指数结果中可见光波段差异植被指数的提取精度最好,能很好地将植被与非植被区分,为荒漠植被覆盖度和植被指数的反演模型的验证提供了依据,节约了实地测量带来的时间和人力成本.对比研究植被指数和荒漠植被覆盖度的线性、对数、三次、乘幂、增长及指数的回归模型结果,最优的荒漠植被覆盖度的反演模型是可见光波段差异植被指数的三次模型,为y=-200.06x3+706.763x2-430.779x+17.916,能很好地监测荒漠植被覆盖度的动态变化,为今后荒漠生态系统的防护和管理提供较好的技术支持.

On the edge of the southeast Tengger Desert,visible-light images were collected by using unmanned aerial vehicle(Phantom 3 advanced).Based on ENVI software,red,green,and blue band information of the photos were analyzed for extraction of vegetation index and vegetation coverage.Regression models were analyzed between five vegetation indices including visible-band difference vegetation index(VDVI),normalized green-red difference index,normalized green-blue difference index,excess green,red-green ratio index and desert vegetation coverage in order to screen out the optimal inversion model to validate the vegetation coverage of unmanned aerial vehicle aerial images from the same area and at the same time.The validity of the inversion models was tested through comparing real vegetation coverage with different vegetation indices.The results showed that,by comparing five vegetation indices that were calculated via visible light band information,VDVI was the best index to distinguish vegetation from non-vegetation and provided a basis for establishing conversion models between vegetation coverage and vegetation indices,which would save the time and manpower cost of field investigation.After comparing linear,logarithmic,cubic,power and exponential regression models,the optimal inversion model was a cubic equation model:y=-200.06x3+706.763x2-430.779x+17.916.The model could be used to monitor the dynamic changes of desert vegetation coverage.In summary,through establishing an inversion model between vegetation coverage and vegetation indexes,we could monitor the dynamics of desert vegetation and provide technical support for the protection and management of the desert ecosystem.