Parcel-level classification of glutinous maize under UAV remote sensing: discriminating laser pretreatment differences

Pre-sowing laser pretreatment can modify maize’s early physiology, but parcel-scale, non-destructive verification across the seed-stage–field-response chain remains limited. We present a UAV multispectral workflow for parcel-level discrimination of glutinous maize subjected to red/blue laser pretreatments. Images are co-registered in a unified coordinate system; vegetation indices (NDVI, NDRE, NDVI705, MTCI, CIre) are computed; and GLCM textures from selected bands enable spectral–texture fusion. Candidate features are ranked using an equal-weight, normalized score that combines ANOVA F, mutual information, and Eta², with a Pearson redundancy constraint (|r|