Mitigating Illumination-, Leaf-, and View-Angle Dependencies in Hyperspectral Imaging Using Polarimetry

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 (ϵ) and a reduction spanning 1.5 to 2.7 in their standard deviation (ϵσ) after applying the correction network on the HIP sensor data.

Hybrid spatial-temporal Mueller matrix imaging spectropolarimeter for high throughput plant phenotyping

Many correlations exist between spectral reflectance or transmission with various phenotypic responses from plants. Of interest to us are metabolic characteristics, namely, how the various polarimetric components of plants may correlate to underlying environmental, metabolic, and genotypic differences among different varieties within a given species, as conducted during large field experimental trials. In this paper, we overview a portable Mueller matrix imaging spectropolarimeter, optimized for field use, by combining a temporal and spatial modulation scheme. Key aspects of the design include minimizing the measurement time while maximizing the signal-to-noise ratio by mitigating systematic error. This was achieved while maintaining an imaging capability across multiple measurement wavelengths, spanning the blue to near-infrared spectral region (405-730 nm). To this end, we present our optimization procedure, simulations, and calibration methods. Validation results, which were taken in redundant and non-redundant measurement configurations, indicated that the polarimeter provides average absolute errors of (5.3±2.2)×10^-3 and (7.1±3.1)×10^-3, respectively. Finally, we provide preliminary field data (depolarization, retardance, and diattenuation) to establish baselines of barren and non-barren Zea maize hybrids (G90 variety), as captured from various leaf and canopy positions during our summer 2022 field experiments. Results indicate that subtle variations in retardance and diattenuation versus leaf canopy position may be present before they are clearly visible in the spectral transmission.

Multistatic fiber-based system for measuring the Mueller matrix bidirectional reflectance distribution function

Bidirectionality effects can be a significant confounding factor when measuring hyperspectral reflectance data. The bidirectional reflectance distribution function (BRDF) can effectively characterize the reflectivity of surfaces to correct remote sensing measurements. However, measuring BRDFs can be time-consuming, especially when collecting Mueller matrix BRDF (mmBRDF) measurements of a surface via conventional goniometric techniques. In this paper, we present a system for collecting mmBRDF measurements using static optical fiber detectors that sample the hemisphere surrounding an object. The entrance to each fiber contains a polarization state analyzer configuration, allowing for the simultaneous acquisition of the Stokes vector intensity components at many altitudinal and azimuthal viewing positions. We describe the setup, calibration, and data processing used for this system and present its performance as applied to mmBRDF measurements of a ground glass diffuser.

Quantification of gray mold infection in lettuce using a bispectral imaging system under laboratory conditions

Gray mold disease caused by the fungus Botrytis cinerea damages many crop hosts worldwide and is responsible for heavy economic losses. Early diagnosis and detection of the disease would allow for more effective crop management practices to prevent outbreaks in field or greenhouse settings. Furthermore, having a simple, non-invasive way to quantify the extent of gray mold disease is important for plant pathologists interested in measuring infection rates. In this paper, we design and build a bispectral imaging system for discriminating between leaf regions infected with gray mold and those that remain unharmed on a lettuce (Lactuca spp.) host. First, we describe a method to select two optimal (high contrast) spectral bands from continuous hyperspectral imagery (450-800 nm). We then explain the process of building a system based on these two spectral bands, located at 540 and 670 nm. The resultant system uses two cameras, with a narrow band-pass spectral filter mounted on each, to measure the bispectral reflectance of a lettuce leaf. The two resulting images are combined using a normalized difference calculation that produces a single image with high contrast between the leaves' infected and healthy regions. A classifier was then created based on the thresholding of single pixel values. We demonstrate that this simple classification produces a true-positive rate of 95.25% with a false-positive rate of 9.316% in laboratory conditions.

Internal defect scanning of sweetpotatoes using interactance spectroscopy

While standard visible-light imaging offers a fast and inexpensive means of quality analysis of horticultural products, it is generally limited to measuring superficial (surface) defects. Using light at longer (near-infrared) or shorter (X-ray) wavelengths enables the detection of superficial tissue bruising and density defects, respectively; however, it does not enable the optical absorption and scattering properties of sub-dermal tissue to be quantified. This paper applies visible and near-infrared interactance spectroscopy to detect internal necrosis in sweetpotatoes and develops a Zemax scattering simulation that models the measured optical signatures for both healthy and necrotic tissue. This study demonstrates that interactance spectroscopy can detect the unique near-infrared optical signatures of necrotic tissues in sweetpotatoes down to a depth of approximately 5±0.5 mm. We anticipate that light scattering measurement methods will represent a significant improvement over the current destructive analysis methods used to assay for internal defects in sweetpotatoes.