Mitigating illumination-, leaf-, and view-angle dependencies in hyperspectral imaging using polarimetry

Authors: KRAFFT, DANNY - North Carolina State University; SCARBORO, CLIFTON - North Carolina State University; HSIEH, WILLIAM - North Carolina State University; DOHERTY, COLLEEN - North Carolina State University; Balint-Kurti, Peter; KUDENOV, MICHAEL - North Carolina State University

Submitted to: Plant Phenomics
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 2/18/2024
Publication Date: 3/22/2024
Citation: Krafft, D., Scarboro, C.G., Hsieh, W., Doherty, C.J., Balint Kurti, P.J., Kudenov, M.W. 2024. Mitigating illumination-, leaf-, and view-angle dependencies in hyperspectral imaging using polarimetry. Plant Phenomics. 6:0157. https://doi.org/10.34133/plantphenomics.0157.
DOI: https://doi.org/10.34133/plantphenomics.0157

Interpretive Summary: Spectral imaging uses multiple bands across the electromagnetic spectrum has been studied extensively in agriculture. While this technology allows us to use data from drones and other types of “remote sensing” systems to evaluate plant health, interpretation of the data is complicated by several factors. Primary among them is the so-called “bidirectional reflectance distribution function” (BRDF) effects, where the glare from the matte or glossy characteristics of a surface or leaf cause light to change its spectrum at different sun- and viewer-angles. This work reports the development of software that uses the polarization state of the reflected light to adjust the perceived color of the leaf to derive a reading closer to its true state. The software reduced errors due to glare about 10-fold compared to the raw readings.

Technical Abstract: Automation of plant phenotyping using data from high-dimensional imaging sensors is on the forefront of agricultural research for its potential to improve seasonal yield by monitoring crop health and accelerating breeding programs. A common challenge when capturing images in the field relates to the spectral reflection of sunlight (glare) from crop leaves that, at certain solar incidences and sensor viewing angles, presents unwanted signals. The research presented here involves the convergence of 2 parallel projects to develop a facile algorithm that can use polarization data to decouple light reflected from the surface of the leaves and light scattered from the leaf’s tissue. The first project is a mast-mounted hyperspectral imaging polarimeter (HIP) that can image a maize field across multiple diurnal cycles throughout a growing season. The second project is a multistatic fiber-based Mueller matrix bidirectional reflectance distribution function (mmBRDF) instrument which measures the polarized light-scattering behavior of individual maize leaves. The mmBRDF data was fitted to an existing model, which outputs parameters that were used to run simulations. The simulated data were then used to train a shallow neural network which works by comparing unpolarized 2-band vegetation index (VI) with linearly polarized data from the low-reflectivity bands of the VI. Using GNDVI and red-edge reflection ratio we saw an improvement of an order of magnitude or more in the mean error and a reduction spanning 1.5 to 2.7 in their standard deviation after applying the correction network on the HIP sensor data.