Combining features selection strategy and features fusion strategy for SPAD estimation of winter wheat based on UAV multispectral imagery

• DOI: 10.3389/fpls.2024.1404238
• Publication Date: 2024-05-10T04:41:20Z
• Authors: Xiangxiang Su; Ying Nian; Hiba Shaghaleh; Amar Hamad; Hu Yue; Yongji Zhu; Jun Li; Weiqiang Wang; Hong Wang; Qiang Ma; Jikai Liu; Xinwei Li; Yousef Alhaj Hamoud
• Keywords: Winter wheat

The Soil Plant Analysis Development (SPAD) is a vital index for evaluating crop nutritional status and serves as an essential parameter characterizing the reproductive growth of winter wheat. Non-destructive accurate monitorin3g wheat SPAD plays crucial role in guiding precise management nutrition. In recent years, spectral saturation problem occurring later stage has become major factor restricting accuracy estimation. Therefore, purpose this study to use features selection strategy optimize sensitive remote sensing information, combined with fusion integrate multiple characteristic features, order improve estimating SPAD. This conducted field experiments different varieties nitrogen treatments, utilized UAV multispectral sensors obtain canopy images during heading, flowering, late filling stages, extracted texture from images, employed (Boruta Recursive Feature Elimination) prioritize features. Support Vector Machine Regression algorithm are applied construct estimation model results showed that NIR band other bands can fully capture differences stage, red more During stability constructed using both superior models only single feature or no strategy. enhancement by method becomes significant, greatest improvement observed R 2 increasing 0.092-0.202, root mean squared error (RMSE) decreasing 0.076-4.916, ratio performance deviation (RPD) 0.237-0.960. conclusion, excellent application potential stages growth, providing theoretical basis technical support precision nutrient crops.