No abstract

A new method called Arbor Crown Enumerator (ACE) was developed for tree crown detection from multispectral Very High-resolution (VHR) satellite imagery. ACE uses a combination of the Red band and Normalized Difference Vegetation Index (NDVI) thresholding, and the Laplacian of the Gaussian (LOG) blob detection method. This method minimizes the detection shortcomings of its individual components and provides a more accurate estimation of the number of tree crowns captured in an image sample. The ACE was applied successfully to sample images taken from a four-band QuickBird (0.7m ϫ 0.7m) scene of Keritis watershed, in the Island of Crete. The method performs very well for different tree types, sizes and densities that may include non vegetation features such as roads and houses. Statistical analysis on the tree crown detection results from the sample images supports the agreement between the measurements and the simulations. The new method reduces considerably the effort of manual tree counting and can be used for environmental applications of fruit orchard, plantation and open forest population monitoring.

No abstract

No abstract

No abstract