Root Detection By Hierarchical Seed Expansion

Analysing plant root images is time consuming but very important for ecological, agronomic and biological studies. In this paper we focus on the second of two phases which expands a given part of a root to cover its whole projection in the image extending our fully automatic analysis approach. Employing a hierarchical A* algorithm allows to use a more sophisticated estimation of the occurring remaining costs thus speeding up search substantially. This approach also allows a finer tuning of the root model and eases the use of previous knowledge by providing an intelligent restriction of the search space