Nondestructive Quantitative Analysis of Water Potential of Tomato Leaves Using Online Hyperspectral Imaging System

Abstract

HighlightsWe developed an online measurement system for water potential of tomato plants using hyperspectral imaging.We used Linear Discriminant Analysis to automatically and quickly extract the leaf images.We used SNV scattering correction to remove the spectral variations caused by collecting the defocused leaf images.We developed a prediction model of leaf water potential based on spectral image information.Abstract. Tomatoes have different water requirements in each growing period. Excessive water use or insufficient water supply will affect the growth and yield of tomato plants. Therefore, precise irrigation control is necessary during cultivation to increase crop productivity. Traditionally, the soil moisture content or leaf water potential has been used as an indicator of plant water status. These methods, however, have limited accuracy and are time-consuming, making it difficult to be put into practice in tomato production. This study developed an online hyperspectral imaging system to measure the leaf water potential of tomato nondestructively. Linear Discriminant Analysis was utilized to automatically and quickly extract the leaf images, with the recognition accuracy of 94.68% was achieved. The mathematical processing of Standard Normal Variate scattering correction was used to remove the spectral variations caused by the defocused leave images. The developed leaf water potential prediction model based on the spectral image information attained using the developed system achieved the standard error of calibration of 0.201, coefficient of determination in calibration set of 0.814 and standard error of cross-validation of 0.230, and one minus the variance ratio of 0.755. The obtained performance indicated the feasibility of applying the developed online hyperspectral imaging system as a real-time non-destructive measurement technique for the leaf water potential of tomato plants. Keywords: Hyperspectral imaging system, Machine learning, Tomato, Water potential.