Abstract
We consider the problem of extracting curve skeletons of three-dimensional,elongated objects given a noisy surface, which has applications in agriculturalcontexts such as extracting the branching structure of plants. We describe anefficient and robust method based on breadth-first search that can determinecurve skeletons in these contexts. Our approach is capable of automaticallydetecting junction points as well as spurious segments and loops. All of thatis accomplished with only one user-adjustable parameter. The run time of ourmethod ranges from hundreds of milliseconds to less than four seconds on large,challenging datasets, which makes it appropriate for situations where real-timedecision making is needed. Experiments on synthetic models as well as on datafrom real world objects, some of which were collected in challenging fieldconditions, show that our approach compares favorably to classical thinningalgorithms as well as to recent contributions to the field.