关键词: 棉花, 无人机遥感, 植被指数, 株高, 叶绿素相对含量, 产量

Abstract: Timely and accurate monitoring of cotton growth and yield is the key to precision farming management. Unmanned Aerial Vehicle (UAV) platforms enable rapid acquisition of remote sensing data with high spatio-temporal resolution, showing great potential in crop growth parameters and yield estimation. Taking cotton in Binzhou City of Shandong Province as the research object, remote sensing images were obtained by using the multi-spectral camera installed on the UAV , and the reflectance of each band was extracted respectively, and 8 vegetation indices were screened out, and three methods such as multiple linear regression (MLR), random forest(RF) and artificial neural network (BPNN) were used to construct estimation models of cotton plant height, relative chlorophyll content and yield per plant respectively, and verified them. The results showed that the accuracy of the inversion model at the mature stage was generally higher than that at the full flowering stage. The R2 of the validation set for cotton plant height estimation in the peak flowering and mature stages were 0.842 and 0.670, respectively. The R2 values for the validation set of the chlorophyll relative content estimation model were 0.725 and 0.765, respectively. The R2 values for the validation set of the yield estimation model were 0.860 and 0.846, respectively. These results provide theoretical basis for the application of UAV remote sensing in crop growth parameters and yield estimation, and also provide a practical reference for further optimization of agricultural production management, scientific decision-making and policy formulation.

Key words: cotton, UAV remote sensing, vegetation index, plant height, SPAD, yield