Extraction of maize vegetation coverage based on UAV multi-spectral remote sensing and pixel dichotomy

Abstract

In order to quickly and accurately obtain the vegetation coverage information at the jointing stage of summer maize, drone remote sensing was used to obtain the multi-spectral images of unmanned farms. Seven multi-spectral vegetation indices were extracted from the images, including normalized vegetation index (NDVI), normalized green belt difference vegetation index (GNDVI)), enhanced vegetation index (EVI), difference vegetation index (DVI), soil adjusted vegetation index (SAVI), ratio vegetation index (RVI), and optimal soil adjusted vegetation index (OSAVI) vegetation coverage extraction model. Using Support Vector Machine supervised classification method was adopted to obtain the true value of vegetation coverage of summer maize jointing period and the calculation accuracy of each vegetation coverage extraction model was evaluated. In order to ensure the accuracy and stability of the extraction results, 50 independent verification units were selected in the experimental areas, and the true value of the vegetation coverage in each area was linearly fitted with the vegetation coverage predicted value obtained by the vegetation coverage extraction model, and each was calculated. The coefficient of determination ( R 2 ) and the root mean square error (RMSE) of the vegetation coverage extraction model, were compared with the fitting accuracy and goodness of fit of the linear regression models, and finally an optimal summer maize jointing stage coverage extraction model was obtained. The results show that the vegetation coverage extraction model constructed based on the pixel dichotomy combined with the multi-spectral vegetation index EVI has the best effect on the extraction of summer maize jointing stage coverage and the highest accuracy. Compared with the true value of vegetation coverage, the extraction error of the EVI vegetation coverage extraction model ( E F ) is 11.76% with R 2 of 0.9418 and RMSE of 0.0537. Based on the pixel dichotomy and using the multi-spectral vegetation index a vegetation coverage model was constructed, and the vegetation coverage of summer maize at the jointing stage can be quickly and accurately extracted. This study can provide technical support for the realization of precision agriculture. Keywords: Unmanned farm, UAV, multispectral remote sensing image, pixel dichotomy, vegetation index, maize vegetation coverage DOI: 10.33440/j.ijpaa.20210402.172 Citation: Yang D J, Lan Y B, Li W H, Hu C X, Xu H Y, Miao J C, Xiao X, Hu L B, Gong D C, Zhao J. Extraction of maize vegetation coverage based on UAV multi-spectral remote sensing and pixel dichotomy. Int J Precis Agric Aviat,2021;4(2): 1–7.