Low-Cost Hyperspectral Imaging to Detect Drought Stress in High-Throughput Phenotyping

Abstract

Recent developments in low-cost imaging hyperspectral cameras have opened up new possibilities for high-throughput phenotyping (HTP), allowing for high-resolution spectral data to be obtained in the visible and near-infrared spectral range. This study presents, for the first time, the integration of a low-cost hyperspectral camera Senop HSC-2 into an HTP platform to evaluate the drought stress resistance and physiological response of four tomato genotypes (770P, 990P, Red Setter and Torremaggiore) during two cycles of well-watered and deficit irrigation. Over 120 gigabytes of hyperspectral data were collected, and an innovative segmentation method able to reduce the hyperspectral dataset by 85.5% was developed and applied. A hyperspectral index (H-index) based on the red-edge slope was selected, and its ability to discriminate stress conditions was compared with three optical indices (OIs) obtained by the HTP platform. The analysis of variance (ANOVA) applied to the OIs and H-index revealed the better capacity of the H-index to describe the dynamic of drought stress trend compared to OIs, especially in the first stress and recovery phases. Selected OIs were instead capable of describing structural changes during plant growth. Finally, the OIs and H-index results have revealed a higher susceptibility to drought stress in 770P and 990P than Red Setter and Torremaggiore genotypes.

Keywords: drought stress; high-throughput phenotyping; hue; hyperspectral index; low-cost hyperspectral camera; optical sensor; projected shoot area; red-edge; senescence index; tomato.

Figure 1

RH expressed in percentage (a), T expressed in °C (b), and VPD expressed in Kilopascal (c) trends during the experiment. The red and blue areas represent the stress and recovery periods respectively. (a,b) report the mean of two sensors and the associated standard deviation. RH data were not recorded on day 31 May.

Figure 2

ET trend by 770P, 990P, Red Setter and Torremaggiore genotypes from the beginning to the end of the experiment. The red line represents the stressed thesis, and the blue line represents the control thesis. The red and blue areas in the plots represent the stress and full-irrigation experimental phases, respectively. Each point reports the mean of six samples and the associated standard deviation.

Figure 3

Example of segmentation results. The segmentation process based on the 96° percentile was applied to a Red Setter tomato genotype control thesis acquired on 8 June 2021. The original HI (a), composed of 1,048,576 pixels, was reduced to 11,408 pixels (b), permitting the sampling of the plant regions characterized by direct and orthogonal illumination.

Figure 4

The spectral signature derived by the selected plant of Figure 3 after segmentation. The red dashed line shows the spectral band (691 nm) where the maximum value of the derivate was detected and used to derive the hyperspectral index (H-index). The blue line reports mean ± standard deviation of 11,408 pixels.

Figure 5

Hyperspectral index (H-index) values in stressed and control treatments (red and blue lines and points, respectively) for the four genotypes. Each point reports the mean of six samples and the associated standard deviation.

Figure 6

Projected shoot areas in stressed and control treatments (red and blue lines and points, respectively) for the four genotypes. Each point reports the mean of six samples and the associated standard deviation.

Figure 7

HUE index values in stressed and control treatments (red and blue lines and points, respectively) for the four genotypes. Each point reports the mean of six samples and the associated standard deviation.

Figure 8

Senescence index values in stressed and control treatments (red and blue lines and points, respectively) for the four genotypes. Each point reports the mean of six samples and the associated standard deviation.

Figure 9

Plants located on the conveyor belt (1) were moved to visible camera (2) during the acquisition process. Subsequently, plants were moved to hyperspectral camera (3) and automatic irrigation system (4). Plants were left on the conveyor belt during the time between acquisition processes. The HTPP was based on a LemnaTec Scanalzer 3D system.

Figure 10

Hyperspectral acquisition process on tomato plant. The white Lambertian surface reference target at 75% of reflectance is the white panel on the plant’s left.