Feature extraction of cotton plant height based on DSM difference method

Abstract

Abstract: Spatiotemporal scale information of cotton plant height is closely related to cotton yield, and can provide a basis for irrigation, fertilization and topping of cotton in the field. In order to realize rapid monitoring of plant height during cotton growth period, this research obtained visible images of cotton initial bud stage, full bud stage and flowering stage by mounting an Unmanned Aerial Vehicle and splices them to obtain its Digital Surface Model. The mean plant height of cotton in each sampling area was calculated by combining the DSM difference method and the sample statistical method of the region of interest to generate a high-level distribution map of cotton plant height with soil background removed. The results showed that the extraction method of plant height of cotton based on DSM difference method is stable ( R 2 of cotton in three periods is higher than 0.750, RMSE is less than 5 cm; plant height model R 2 =0.849 of cotton in the whole growth period). Therefore, DSM difference method can be used to achieve rapid and effective acquisition of cotton plant height, which is of great significance to guiding field cotton management. Finally, the vector files of pure cotton vegetation generated by the cotton TRVI vegetation index threshold method were used to crop the cotton DSM difference images, retaining the areas of only cotton vegetation in the test area, and generating the plant height class distribution maps of cotton at the first bud, full bud and flowering stages, so as to provide a theoretical basis for topping, irrigation and fertilization operations of cotton in the field. Keywords: visible images; unmanned aerial vehicle; plant height; digital surface model DOI: 10.33440/j.ijpaa.20210401.151 Citation: Yang H B, Hu X, Zhao J, Hu Y H. Feature extraction of cotton plant height based on DSM difference method. Int J Precis Agric Aviat, 2021; 4(1): 59–69.