Multi-scale reconstruction of crop canopy

Abstract

We present methods for three-dimensional (3D) reconstruction of a crop canopy from multi-scale measured data and a method for verifying the accuracy of these reconstructed canopies. The validation method aims to give a quantitative evaluation to these reconstruction methods and reconstructed canopies, because different reconstruction methods use different data and also involve different time and labor costs for measuring data from real crops. Our reconstruction methods include reconstruction from digitized feature points of the crops, reconstruction by combining digitized feature points and scanned leaf mesh, and reconstruction from statistical models of morphological structure about the crops. Projected area (from different viewpoints) of the reconstructed crops 3D canopy was used as the key factor for verifying the accuracy of the reconstructed models. We concluded that reconstruction from digitizing characteristic points and scanned leaf mesh will obtain favorable result with low time and labor costs for measuring data.