An Image-based Canopy Reconstruction Framework for Field Maize

Abstract

Plant canopy reconstruction is the fundamental and central problem of functional-structural plant modeling. In order to reconstruct geometric models of plant canopy by real-time image driving, this paper proposes a framework of imagebased maize canopy modeling. The framework builds field image and meteorological data acquisition system to acquire real-time information of plant canopy, then extracts plant growth position and plant azimuthal plane of each plant in the canopy. Morphological parameters of plants are generated using crop model. Finally, geometric model of target canopies are reconstructed by parametric modeling and self-regulating from the image extracted coverage parameter. Experimental results shows that the average coverage error of five treatment maize canopies is 2.57% and LAI errors are smaller than ±1%. Light distribution verification average error of five canopies is 8.02%. This framework will provide technical support for maize field growth analysis and decision making based on Internet of things.