Abstract:Chlorophyll is an important pigment in photosynthesis of plants. It can provide important basis for realtime monitoring and health diagnosis of crops by using crop spectral and texture information to retrieve chlorophyll. In the field environment, five different varieties of corn at four leaves stage and jointing stage were selected as the research objects. The visible light images were obtained by UAV, and the soil background was dealt with mask treatment. Totally 25 kinds of visible light vegetation index and 24 kinds of texture features were extracted. The correlation between vegetation index, texture feature and relative chlorophyll content (SPAD) of corn was comprehensively analyzed, and stepwise regression (SR), partial least squares regression (PLSR) and support vector regression (SVR) models based on vegetation index, texture feature and vegetation index + texture feature were respectively established to quantitatively estimate relative chlorophyll content. In SR model, vegetation index+texture feature model was the same as vegetation index model, which was better than texture feature model, R2 was 0.7316, RMSE was 2.9580, RPD was 1926; in PLSR model, vegetation index+texture feature model was better, texture feature model was the second, vegetation index model was the worst, R2 was 0.8025, RMSE was 2.4952, RPD was 2.284; in SVR model, vegetation index + texture feature model was better, vegetation index model was next, texture feature model was the worst, R2 was 0.8055, RMSE was 2.6408, RPD was 2158. Comprehensive analysis using the PLSRbased vegetation index+texture feature model can achieve rapid and accurate extraction of corn SPAD, providing a new method and experience for chlorophyll inversion, and also providing a reference for UAV remote sensing growth monitoring.