Plant Hyperspectral Imaging

Abstract

Hyperspectral imaging can generate spatial chemical information in plants. The imaging acquisition system is basically composed of a radiation source, sample stage, objective lens, spectrograph, CCD camera and a computer to store and process derived data. Most hyperspectral imaging acquisition approaches are nondestructive in nature and require minimum sample preparation, thus producing chemically rich information without modifying a sample's features. Data processing is mainly performed via multivariate image analysis (MIA), where computed‐based methods are employed for preprocessing, feature extraction and multivariate analysis towards classification. Applications vary according to the desired information of interest, but they mainly include textural analysis, chemical and biochemical analysis and plant disease identification. Successful studies in these areas reinforce the sensitivity and versatility of hyperspectral imaging in plants. Key Concepts Hyperspectral imaging is a powerful tool for analysing plants. Both spatial and spectral information are acquired via hyperspectral imaging, generating chemically rich information with spatial distribution. Most hyperspectral imaging techniques are nondestructive in nature and require minimal sample preparation. Data are processed by computational‐based methods, in particular, by using multivariate image analysis (MIA) techniques for feature extraction and classification. Hyperspectral imaging is a versatile analytical technique with many applications in plant studies, such as textural analysis, chemical and biochemical analysis and disease identification.

Hyperspectral imaging for early detection of foliar fungal diseases on small grain cereals: a minireview

“…The hyperspectral curve obtained by measuring plant leaves contains noise and variations of structural features, so there is a need to do some curve transformations that should be performed before spectral analysis to describe the spectral curve more accurately and according to the most important structural features (Song et al, 2011). Generally, information gathering from hyperspectral images requires the following steps (methods): (i) preprocessing treatment, (ii) feature extraction, (iii) analysis, and (iv) acquisition of desired information (Bravo et al, 2003;Morais et al, 2019).…”

“…(Lu et al, 2020). A hyperspectral imaging system consists of four main elements: (i) a camera, (ii) a light source, (iii) a sample stage (no obligate), and (iv) corresponding control software (Morais et al, 2019). The main elements of the imaging unit are the lenses, an imaging spectrograph, and an area detector.…”

“…The scattered light is projected onto a detector that converts the photons into electrical signals that can be measured as the strength of different wavelengths into intensity values. Mostly there are two types of detectors -charge-coupled-device (CCD) camera or complementary metal-oxide-semiconductor (CMOS) camera (Mishra et al, 2017;Morais et al, 2019).…”