Multi-Model Rice Canopy Chlorophyll Content Inversion Based on UAV Hyperspectral Images

Rice is China’s main crop and its output accounts for 30% of the world’s total annual rice production. growth status closely related to chlorophyll content (called Soil Plant Analyzer Development (SPAD) values). The determination a SPAD value great significance health rice, agricultural irrigation regulated fertilization. traditional measurement method not only time-consuming, laborious expensive but also causes irreparable damage vegetation. aim present study obtain through inversion hyperspectral remote sensing images. In order achieve this purpose, image at different stages canopy scale was first acquired using imaging instrument equipped with drone; spectral characteristics were analyzed combined ground-level measured value, bands high correlation between values spectra fertility selected. Subsequently, we continuous projection algorithm extract characteristic band used PLS in MATLAB software analyze calculate weight each type corresponding value; then wavelength highest as band. Secondly, four methods univariate regression, partial least squares (PLS) support vector machine (SVM) regression back propagation (BP) neural network are integrated establish estimation model canopy. Finally, models map Research shows that decision coefficient among booting stage “booting stage-SVR” (R2 = 0.6258), dairy maturity “milk-ripe stage-BP” 0.6716), all which can meet requirement accurately retrieving above results provide technical reference accurate, rapid non-destructive monitoring leaves core selection basis large-scale rice.