Reconstructing tree trunks from point clouds using PSO

Trees have a common tendency to grow vertically and bent trees usually imply previous movements of underlying soils. For trees located on a landslide site, the tilt and bend angles of the trees can therefore be used to estimate the sliding distance of the slope. To achieve this end and to serve as a pilot study, this paper studied the feasibility of scanning trees using a terrestrial laser scanner and reconstructing the tree trunks by fitting cylinders to the resulting point clouds using the particle swarm optimization(PSO) technique. The results were very satisfactory and a parametric study was done to demonstrate the influence of parameter selection.