Segmentation and geometry identification of white roots in two-dimensional scanner images

This paper describes the concept of segmentation and geometry restauration for two-dimensional scanner images of white roots from hydroponic cultures. The segmentation applies RGB to HSV colour spaces transform and takes into account high chroma and dark type discolourations of root tissue. To identify spatial root tree, all paths in its 2D projection graph are individually tracked, with some a priori assumptions about the tree. They enable the detection of lateral root intersections and the elimination of redundant path links after tracking.