Prediction of Seed Vigour and Viability Using Hyperspectral Imaging
Neeraja Rajan *
Department of Seed Science and Technology, Kerala Agriculture University, Vellanikara, Kerala, India.
Vidhu Francis Palathingal
Department of Seed Science and Technology, Kerala Agriculture University, Vellanikara, Kerala, India.
Dijee Bastian
Department of Seed Science and Technology, Kerala Agriculture University, Vellanikara, Kerala, India.
Aparna A Raj
Department of Seed Science and Technology, Kerala Agriculture University, Vellanikara, Kerala, India.
Monika K. G
Department of Seed Science and Technology, Kerala Agriculture University, Vellanikara, Kerala, India.
S. Deepthy
Department of Seed Science and Technology, Kerala Agriculture University, Vellanikara, Kerala, India.
*Author to whom correspondence should be addressed.
Abstract
The present review outlines the prospects for Hyperspectral imaging (HSI) implementation as a standard tool for predicting seed vigour, viability in commercial agriculture. The most extensively used optical detection methods for seed vitality mainly include machine vision detection, near-infrared spectroscopy, hyperspectral detection, Raman spectroscopy, fluorescence spectroscopy, and seed exhalation gas spectroscopy. Seed vigour and viability are integral to the performance of crops, from germination rates and seedling establishment to final yield potential. Traditional methods for assessing seed vigour and viability outcomes are reliant on time, labour, or both methods, which often result in destructive sampling of the seeds, making them cumbersome and unsuitable for high-volume seed quality assessment. Hyperspectral imaging (HSI) is a new and fast analytical method which is non-destructive to the sample and has great potential to estimate seed vigour and viability. The reviews identify how both spectral features and spatial features can be combined to monitor subtle biochemical and structural differences of seeds regardless of whether the seeds are affected by ageing, mechanical damage or physiological stress. The review also presents HSI results that have spectral regions that can differentiate germinable seeds and non-germinable seeds, machine learning algorithms capabilities to improve yield, and the use of high-throughput systems with faster and real-time data presentations. The potential for HSI to be used in seed science is enormous and creates a transformational opportunity for more rapid, accurate, and non-destructive seed testing for vigour and viability.
Keywords: Hyperspectral imaging, seed germination, non-destructive testing, spectral analysis, machine learning algorithms