A new sigmoid-based nonlinear vegetation index for tree canopy extraction from high-resolution satellite images

This study develops an innovative sigmoid-based vegetation index, Sigmoid-based Tree Canopy Index (STCI), for tree canopy extraction, aiming to effectively separate tree canopies from surrounding vegetation. STCI enhances vegetation index through a targeted sigmoid transformation to selectively amplify canopy-structure-related spectral signals. The results demonstrate that STCI exhibits strong robustness across different land-cover types and optical sensors, showing stable performance comparable to traditional indices, particularly within urban forest environments. Furthermore, STCI not only distinguishes tree canopy and other vegetation but also improves vegetation classification accuracy, achieving an enhancement of canopy classification accuracy approximately 1–13% compared with traditional indices. This improvement remains consistent cover whole Taiwan, as well as across different seasons, indicating strong spatial–temporal robustness. STCI was also tested on multiple high-resolution satellite images from different sensors, maintaining good applicability despite minor reductions in performance due to differences in spatial resolution and sensor-specific spectral responses. In addition, both the cross-sensor evaluation and the equation-level analysis highlight the essential role of the blue band in distinguishing tree canopies from adjacent vegetation. Overall, this study introduces a simple in vegetation index design and demonstrates that incorporating band-ratio-based sigmoid functions can substantially enhance tree canopy extraction.