Automated threshold extraction and dynamic monitoring of built-up areas using multi-indices and hybrid optimization algorithms

Efficient and accurate extraction of urban built-up areas (BUA) is essential for analyzing spatiotemporal dynamics of human activities and assessing their environmental impacts. However, existing methods face notable limitations: classification-based approaches require large training datasets and show limited cross-temporal adaptability, whereas index-based thresholding often relies on subjective settings and external references, reducing scientific rigor and generalizability. To overcome these challenges, this study proposes a novel framework for automatic BUA extraction, which is applied to China from 2002 to 2022. A Multi-data Adjusted Nighttime Light Urban Index (MANUI) has been developed to enhance feature contrast between BUA and non-BUA. Furthermore, an alternating hybrid Genetic Algorithm and Particle Swarm Optimization approach is applied to optimize multi-threshold OTSU segmentation, enabling objective and adaptive threshold determination. Validation across Chinese cities shows a strong correlation with visually interpreted data (R² = 0.94) and a relative error of 21.69%, outperforming existing datasets. Using this framework, spatiotemporal patterns of urban expansion and agglomeration were examined. The results indicate continuous growth in eastern and southern coastal regions, alongside an emerging trend toward more balanced inland development. Overall, the proposed framework provides a robust tool for dynamic urbanization monitoring, with practical implications for urban planning and land-use management.