Dense 3D datasets for plant modelling

Speaker: Andrew French

Abstract: Previously we have presented work on phenotyping plants, specifically capturing traits such as sizes and shapes of roots. Our recent work has been focused on canopies, for which we use 3D capture techniques. The resulting dataset is a large (millions) point dataset from which we need to make sense of individual plants and measurements such as leaf orientation, surface area etc. This becomes a general surface fitting problem to large 3D datasets. This talk presents an overview of how to manage the dataset: how to capture it, and how to process it afterwards.