Automated recovery of three-dimensional models of plant shoots from multiple color images.

Michael P Pound,Tony P Pridmore,Andrew P French,Erik H Murchie

doi:10.1104/pp.114.248971

Abstract

Increased adoption of the systems approach to biological research has focused attention on the use of quantitative models of biological objects. This includes a need for realistic three-dimensional (3D) representations of plant shoots for quantification and modeling. Previous limitations in single-view or multiple-view stereo algorithms have led to a reliance on volumetric methods or expensive hardware to record plant structure. We present a fully automatic approach to image-based 3D plant reconstruction that can be achieved using a single low-cost camera. The reconstructed plants are represented as a series of small planar sections that together model the more complex architecture of the leaf surfaces. The boundary of each leaf patch is refined using the level-set method, optimizing the model based on image information, curvature constraints, and the position of neighboring surfaces. The reconstruction process makes few assumptions about the nature of the plant material being reconstructed and, as such, is applicable to a wide variety of plant species and topologies and can be extended to canopy-scale imaging. We demonstrate the effectiveness of our approach on data sets of wheat (Triticum aestivum) and rice (Oryza sativa) plants as well as a unique virtual data set that allows us to compute quantitative measures of reconstruction accuracy. The output is a 3D mesh structure that is suitable for modeling applications in a format that can be imported in the majority of 3D graphics and software packages.

Highlights

Increased adoption of the systems approach to biological research has focused attention on the use of quantitative models of biological objects
Verification of our approach is achieved using a unique artificial data set in which an in silico model rice plant is rendered from multiple viewpoints to generate artificial color images that are treated in the same way as a real-world image set
We have tested our reconstruction methods on data sets obtained from rice (Oryza sativa) and wheat (Triticum aestivum) plants

Summary

Introduction

Increased adoption of the systems approach to biological research has focused attention on the use of quantitative models of biological objects This includes a need for realistic three-dimensional (3D) representations of plant shoots for quantification and modeling. 3D geometrical models contain the information needed to compute summary plant traits, such as total leaf area, mean leaf angle, etc These underpin both plant breeding programs and attempts to understand the relationship between genotype, phenotype, and environment, regardless of the scientific approach taken. Depending on the plant species, leaves can lack the texture necessary to perform robust feature matching, either to separate leaves from one another or to locate specific leaves across multiple views To overcome this challenge, where image-based modeling approaches are successful, they have often involved user interaction to guide the process (Quan et al, 2006)

Objectives

Methods

Results

Conclusion