Unmanned Aerial Vehicle-Based Hyperspectral Imaging Integrated with a Data Cleaning Strategy for Detection of Corn Canopy Biomass, Chlorophyll, and Nitrogen Contents at Plant Scale

• DOI: 10.3390/rs17050895
• Publication Date: 2025-03-04T10:41:51Z
• Authors: Zhuolin Shi; Linglong Wang; Zengling Yang; Jinzhao Li; Linwei Cai; Yuanping Huang; Hongyan Zhang; Lujia Han

The high-frequency detection of plant-scale crop growth in the field has great significance for achieving precise crop management and improving breeding practices. In this study, the biomass (BM), chlorophyll (Chl), and total nitrogen (TN) contents of the upper three leaves of the corn canopy are taken as examples, and unmanned aerial vehicle (UAV) and indoor hyperspectral imaging (HSI) detection models are established using partial least squares regression and support vector machine regression, respectively. The performance of the UAV HSI model was notably lower in comparison to the indoor model. Therefore, a UAV HSI data cleaning strategy integrated with RGB image information is further proposed, which involves eliminating data points with serious interference from information non-related to the plant. After data cleaning, the R2C of the BM, Chl, and TN contents detected through UAV HSI reached 0.537, 0.852, and 0.657, representing an improvement of over 70%. The RMSEP values were as low as 0.50 g, 2.2 SPAD, and 0.258%, which were comparable to those obtained with the indoor HSI detection model. This study demonstrates that UAV HSI integrated with the proposed data cleaning strategy can enable the rapid detection of corn canopy leaf properties at the plant scale in the field, supporting the high-frequency characterization of plant-scale crop growth parameters in the field.