Research on the extraction of vegetation coverage in arid desert area based on visible light images

Objective In order to quickly and accurately obtain the information of vegetation coverage in arid desert area. Method In this study, based on the visible light images collected by UAV ( Phantom 4 Advanced ) near Table Mountain in Haibowan District, Wuhai City, Inner Mongolia, the segmentation threshold was determined by the combination of supervised classification and vegetation index statistical histogram, and the vegetation information obtained by visual interpretation was used as the true value for accuracy verification. Results The results show that : ( 1 ) When the visible light vegetation index is used to extract the vegetation information in the study area, the pixel statistical histogram is difficult to present the bimodal characteristics. There is a large overlap between the vegetation and the bare land pixel values in the gray images of the normalized green-red difference index ( NGRDI ), red-green ratio index ( RGRI ) and vegetation index ( VEG ). ( 2 ) The combination of supervised classification and vegetation index statistical histogram can well solve the problem that the vegetation index histogram method is difficult to show bimodal characteristics when it is applied to vegetation extraction in arid areas, and can improve the extraction accuracy of vegetation. ( 3 ) Through the verification of the extraction accuracy of each vegetation index, the difference enhanced vegetation index ( DEVI ) is better than other vegetation indexes in the study area, and the accuracy can reach 96.44 %, and the threshold stability is better. Conclusion The method of combining supervised classification with vegetation index histogram to determine the threshold has a good effect on extracting vegetation information in arid desert area, which has practical significance for the application of visible light image vegetation information extraction in arid desert area.