利用无人机多光谱影像分类电力廊道树种OACSTPCD

Bamboo and eucalyptus have become one of the main tree barrier threats faced by high-voltage transmission lines in densely vegetated areas due to their fast growth and high canopy.The identification of bamboo and eucalyptus is the key to early tree barrier warning.We used the multi-rotor UAV to obtain the multi-spectral images of high-voltage transmission corridor area,and made full use of the advantages of red edge band to extract the spectral features and vegetation index features at first.And then,we used the random forest(RF)algorithm to optimize the vegetation index features.Finally,we compared the maximum likelihood and support vector machine classification methods.Feature optimiza-tion based on RF can not only reduce the dimension of feature variables,but also maintain the original classification accuracy,so as to improve the efficiency of classification.