High-throughput estimation of plant height and above-ground biomass of cotton using digital image analysis and Canopeo

Plant height and above-ground biomass are important growth parameters that affect crop yield. Efficient and non-destructive technologies of crop phenotypic monitoring play crucial roles in intelligent farmland management. However, the feasibility of using these technologies to estimate cotton plant height and above-ground biomass has not been determined. This study proposed a low cost and high-throughput imaging method combined with Canopeo to extract the percentages of green color from high-definition digital images and establish a model to estimate the cotton plant height and above-ground biomass. The plant height and above-ground biomass field trials were conducted at two levels of irrigation (soil water content 70% ± 5% and 40%−45%, respectively) using 80 cotton genotypes. The linear fitting performed well across the different cotton genotypes (PH, R2 = 0.9829; RMSE = 2.4 cm; NRMSE = 11% and AGB, R2 = 0.9609; RMSE = 0.6 g / plant; and NRMSE = 5%), and two levels of irrigation (PH, R2 = 0.9604; RMSE = 2.15 cm; NRMSE = 6% and AGB, R2 = 0.9650; RMSE = 4.51 g/plant; and NRMSE = 17%). All reached a higher fitting degree. Additionally, the most comprehensive model to estimate the cotton plant height and above-ground biomass (Y = 0.4832*X + 11.04; Y = 0.4621*X − 0.3591) was determined using a simple linear regression modeling method. The percentages of green color positively correlated with plant height and above-ground biomass, and each model exhibited higher accuracy (R2 ≥ 0.8392, RMSE ≤ 0.0158, NRMSE ≤ 0.06%). Combining a high-definition digital camera with Canopeo enables the prediction of crop growth in the field. The simple linear regression modeling method and the most comprehensive model enable the rapid estimation of the cotton plant height and above-ground biomass. This method can also be used as a baseline to measure other important crop phenotypes.