Automated individual tree-crown delineation and treetop detection with very-high-resolution aerial imagery

Forest stands, as the basic units in forest management, play a pivotal role in understanding the function and service of the forest system. A stand is a contiguous area that contains a number of trees that are relatively homogeneous or similar in species composition or age and different from adjacent areas (Lindenmayer and Franklin 2002). Several parameters of the stand are of particular interest to foresters, including tree density, stand basal area, stand diameter, stand height, crown closure, stand volume, stand table, and site index. Traditionally, to acquire those parameters, field plots with a random, stratified, or systematical sampling scheme have to be designed and measured, which is usually expensive and labor-intensive. Nevertheless, timely and accurately obtaining the stand information is critically important for updating the forest inventory (Spurr 1948) and for conducting ecological studies with those parameters as the input (Palace et al. 2007). As remote sensing imagery is more readily accessible, information gathering becomes more frequent and cost-effective.