author = {Will Gittoes and Tom Botterill and Richard Green},
title = {Quantitative analysis of skeletonisation algorithms for modelling of branches},
booktitle = {In Proceedings of Image and Vision Computing New Zealand},
year = {2011}, abstract={Skeletonisation is an important low-level problem in
computer vision with many applications in shape finding,
motion tracking, character recognition and segmentation. This
paper examines how skeletonisation can be used to find and
model the path of plant branches in an image. A proposed
method for quantitatively comparing the accuracy of skeletons
is described, which compares a skeleton produced by a
skeletonisation algorithm to a ground truth. This method is
used to evaluate several skeletonisation algorithms within the
context of branch modelling. The best single skeletonisation
method is found to be morphological thinning, due to the
highly connected nature of the skeleton.}

Home - Photos & Trip Reports - Tom's research - Hil's Hydrology research