High throughput in vivo analysis of plant leaf chemical properties using hyperspectral imaging

Authors

Piyush Pandey, University of Nebraska-LincolnFollow
Yufeng Ge, University of Nebraska-LincolnFollow
Vincent Stoerger, University of Nebraska-LincolnFollow
James C. Schnable, University of Nebraska-LincolnFollow

ORCID IDs

James C. Schnable

Date of this Version

4-2017

Citation

Poster Presented at University of Nebraska-Lincoln Research Fair, April 4-5, 2017

Comments

Copyright (c) 2017 Piyush Pandey, Yufeng Ge, Vincent Stoerger, James Schnable

Abstract

The possibility of predicting plant leaf chemical properties using hyperspectral images was studied. Sixty maize and 60 soybean plants were used, and two experiments were conducted: one with water limitation and the second with nutrient limitation, with the purpose of creating wide ranges of these chemical properties in plant leaf tissues. A hyperspectral imaging system with a spectral range from 550 to 1700 nm was used to acquire plant images in a high throughput fashion (plants placed on an automated conveyor belt). Leaf chemical properties were measured in the laboratory. Partial least squares regression was implemented on spectral data to successfully model and predict water content, micronutrient, and macronutrient concentrations.