Intelligent evaluation of ornamental characteristics of garden chrysanthemums using unmanned aerial vehicle (UAV)-based RGB imagery

Garden chrysanthemums (Chrysanthemum × morifolium Ramat.) are widely appreciated for their rich diversity in color, form, and blooming period. However, traditional evaluations of ornamental traits rely on subjective and labor-intensive methods, limiting efficiency and consistency. To address this, we developed an UAV-based phenotyping framework to objectively assess ornamental traits in chrysanthemums. Using the UAV-based RGB imaging platform, we evaluated forty cultivars and extracted eleven ornamental traits, including plant height, canopy orthophoto area, canopy roundness, clump cohesion, ornamental duration, flowering rarity grade, flowering efficiency index, visual uniformity, flower color coverage, color saturation, and color value. Principal component analysis (PCA) reduced trait dimensionality, with four principal components explaining 77.94 % of total variance. K-means clustering grouped cultivars into three performance categories (excellent, moderate, poor), supported by hierarchical clustering. Color group analysis showed that pink-purple cultivars exhibited the greatest color diversity, while red cultivars were more uniform. Traits associated with color and texture (e.g., color coverage, saturation, visual uniformity) and floral duration (e.g., efficiency index, ornamental duration) were most influential in cultivar differentiation. Twelve cultivars, such as ‘Jinling Guili’, were identified as excellent based on their visual appeal and landscape suitability. Our results demonstrate that UAV-based phenotyping provides an efficient, objective, and scalable method for cultivar evaluation. The proposed framework identifies color–texture and floral duration as core drivers and offers valuable tools for chrysanthemum breeding and landscape application. More broadly, this study demonstrates the potential of integrating remote sensing and multivariate analysis to modernize ornamental plant evaluation across diverse germplasm resources.