Computational style quantification and spatiotemporal evolution of longmen farmer paintings via feature fusion

This study proposes a computational framework for quantifying the visual style and temporal evolution of Longmen Farmer Paintings through color–texture feature fusion. A dataset of 150 representative works from 1980 to 2024 was constructed, and HSV color moments, GLCM texture descriptors, and dominant color features were extracted for K-means + + clustering. The results identified five stylistic clusters and revealed an observable redistribution from Retro Rustic features toward Modern Decorative tendencies across the defined time windows. AHP-TOPSIS evaluation further showed that the computation-driven design scheme achieved a higher relative closeness score than the experience-driven baseline, particularly in cultural semantic fidelity and texture consistency. The framework provides an interpretable and structured method that may support the quantitative analysis of selected visual features and provide references for heritage-related design translation.