A Time Series Dataset of NIR Spectra and RGB and NIR-HSI Images of the Barley Germination Process

TL;DR

This work presents an open-source time-series dataset of 2242 malting barley kernels imaged with RGB and NIR-HSI cameras, capturing pre-moisture and daily post-moisture states over five days with per-kernel germination labels. It details a robust, cross-modal data collection and processing pipeline that uses ArUco-based grid tracking for RGB and a chessboard-based approach for HSI to localize kernels, and computes mean pseudo-absorbance spectra from segmentation masks. The dataset, including segmentation masks, NIR spectra, and grid coordinates, enables multi-modal analysis of germination dynamics and supports threshold-based segmentation and spectral analyses. While germination was impeded by glue residues from 3D-printed grid components, the resource remains valuable for classification, time-series modeling, and transferable germination studies in barley and related cereals.

Abstract

We provide an open-source dataset of RGB and NIR-HSI (near-infrared hyperspectral imaging) images with associated segmentation masks and NIR spectra of 2242 individual malting barley kernels. We imaged every kernel pre-exposure to moisture and every 24 hours after exposure to moisture for five consecutive days. Every barley kernel was labeled as germinated or not germinated during each image acquisition. The barley kernels were imaged with black filter paper as the background, facilitating straight-forward intensity threshold-based segmentation, e.g., by Otsu's method. This dataset facilitates time series analysis of germination time for barley kernels using either RGB image analysis, NIR spectral analysis, NIR-HSI analysis, or a combination hereof.

Figures (17)

• Figure 1: A Petri dish imaged at day 1. The barley kernel inside the top left square is the one shown in Table \ref{['tab:grain_evolution']}.
• Figure 2: Histograms showing how the number of germinated kernels evolves from day to day.
• Figure 3: Illustration of the imaging setup and example raw camera outputs.
• Figure 4: An overview of the image processing pipeline using Petri dish 26 as an example. Notice how the grid corners are uniquely defined by the detection in the RGB image in Figure \ref{['fig:reference_grid_rgb']}. Localization of the same grid in a subsequent RGB or HSI image can be derived based on corresponding affine transformation matrices based on the ArUco codes' corners and the chessboards' centers, respectively.
• Figure 5: An overview of the image processing pipeline using Petri dish 26 as an example. Notice how the grid coordinates are uniquely defined by the detection in the RGB image in Figure \ref{['fig:reference_grid_rgb']}. Localization of the same grid in a subsequent RGB or HSI image can be derived based on corresponding affine transformation matrices based on the ArUco codes' corners and the chessboards' centers, respectively.