Mapping of African urban settlements using Google Earth images

The main purpose of this work is to explore a reproducible way of replacing manual digitalizations of urban agglomerations using textural image processing from free high-resolution Google Earth images. Photo-interpretation from very high-resolution satellite images is a reliable method for extracting urban agglomerations, but it is extremely time consuming. As an alternative to manual digitalization, a numerical image analysis method based on mathematical morphology has been developed to detect textural patterns associated with built-up areas. Hundreds of urban agglomerations were extracted using this method and then compared to an open geodatabase. The results were better for humid regions, and omission errors were consistently larger than commission errors for the whole dataset, underlining the challenge of detecting highly scattered habitats with texture analyses. However, the method could provide an appropriate estimate of overall urbanization. This research highlights the weaknesses and the strengths of the method compared to manual digitalization, illuminating paths for future improvement of alternative procedures to manual digitalization.

Automated summary

This research explores a reproducible method for replacing manual digitalizations of urban agglomerations using numerical image analysis, which provided better results in humid regions but faced challenges in detecting highly scattered habitats with texture analyses.