Vulnerable land ecosystems classification using spatial context and spectral indices

A novel object-based classification approach combining spectral indices and spatial context achieved the highest accuracy (90.07%) for mapping vulnerable ecosystems in Teide National Park, Spain. The study leveraged WorldView-2 imagery, integrating two sensor-specific indices (WorldView Vegetation Index and Soil Index) with Grey Level Co-occurrence Matrices (GLCM)-dissimilarity to improve habitat classification precision. This method outperformed traditional pixel-based techniques by incorporating both spectral and spatial information, offering a robust tool for conservation monitoring in high-pressure natural areas.

Ibarrola-Ulzurrun, E., Gonzalo-Martín, C., & Marcello-Ruiz, F. J. (2017). Vulnerable land ecosystems classification using spatial context and spectral indices. Earth Resources and Environmental Remote Sensing/GIS Applications VIII, 29. https://doi.org/10.1117/12.2278496