Delineation and identification of individual trees in the Eastern Deciduous Forest

The Ecological Evaluation using Remote Sensing (EERS) group at West Virgini a University is studying the health and status of West Virginia's forests u sing high spatial resolution imagery. Central to our work is a focus on cla ssification and mapping of trees. This paper reports on our initial finding s regarding the delineation of individual trees, and discusses future direc tions we hope to pursue. In a separate paper (Key et al, in this volume) we discuss tree species classification using multi-temporal imagery. Delineation of individual trees in the Eastern Deciduous Forest is challeng ing due to the variety of scales of tree canopy size, the relatively flat t opography of the canopy, and the complex mosaic of the individual crowns. N evertheless, the shadows between crowns provide a good first cut for identi fying tree boundaries. A rank normalization is required to reduce problems due to variable illumination and vignetting. The size of the moving window used in this normalization is crucial in determining the scale of shadows t hat are enhanced. A window approximately the size of the average tree tends to enhance branching within the crown, whereas a window approximately thre e times the size of the average tree enhances individual tree crowns. The s hadows are, however, in short, separate segments that do not isolate the tr ees. These segments can be connected by orientation information obtained fr om a direction of minimum texture algorithm. For each pixel in the image, t exture is calculated over narrow groups of pixels (1 pixel wide by 11 long) centered on the pixel of interest. The orientation of these groups is incr emented by a small angle until all directions have been tested. The directi on with the lowest texture is written out to a new file. A rule-based algor ithm is currently being developed to use this information to join shadow se gments.