Tree extraction from a Very High Resolution Aerial Image by information fusion

Accurate location of tree regions is valuable for a wide range of applications related to urban development planning and land management. The present study explores the benefits of combining colour and texture information in a true colour aerial image for tree detection. A high level fusion scheme was employed for the inversion of a canopy reflectance model using Very High Resolution (VHR.) Aerial Images. Being completely automated and image-based and independent on extensive and impractical surface measurements, the retrieval scheme has potential for operational use and can quite easily be implemented for other regions. More validation studies are needed to evaluate the usefulness and limitations of the approach for other environments and species compositions.