Machine Learning-Based Estimation Of Potato Chlorophyll Content At Different Growth Stages Using Uav Hyperspectral Data

Accurate estimation of chlorophyll (Chl) content is highly significant in monitoring potato growth and improving yield and quality. Fractional differentiation can refine the local information of the spectrum and is conducive to the removal of background noise. In this study, a new method to examine the effects of fractional differentiation on the estimation of the Chl content of crops was developed. Potato (Solanum tuberosum L.) was selected as the research object. A fractional derivative was used for unmanned aerial vehicle (UAV) hyperspectral data processing, and an algorithm for estimating the potato Chl content was studied.The results concluded that the correlation increased after first declining with increasing differential order; the maximum absolute values of the correlation coefficient at different stages were obtained with 1-order differentiation at the budding stage, 0.6-order differentiation during the tuber formation and tuber growth stages and 1.2-order differentiation at the starch accumulation stage. The comparison and analysis of the estimation models of the potato Chl content at different growth stages showed that the support vector machine (SVM) model had the greatest accuracy in estimating the potato Chl content with an R-2 value of 0.83 at the budding stage, followed by R-2 of 0.80 at the tuber forming stage.