Hyperspectral imaging is one of the emerging fields and has advantage of optical spectroscopy as an analytical tool combined with dimensional object visualization obtained by optical imaging. It is differentiated from multi-spectral imaging (MSI) by resolution as MSI collects images from specific areas of the electromagnetic spectrum having imaging from 4 to 20 colours channel. HIS consist of more than 20 bands of spectral data for each pixel. It offers multiple advantages such as higher image acquisition, higher specificity and granularity, great speed, and non-invasive imaging. It has developed drastically from a remote-sensing satellite into a rugged compact, complex, with economically priced imaging and spectroscopic tool for a range of monitoring and diagnostics application.
Hyperspectral imaging is widely used by defence, government agencies, and non-military sectors such as agriculture and mineral mapping. In agriculture HSI systems are widely used to measure airborne crop measurement that involves a hyperspectral imaging system flying over the field where the agrarian economist compares the data collected and estimate the upcoming harvest accurately.
Stress detection: The identification of plant diseases and plant stress is one of the major challenges for agrarian and the existing methods often rely on checking the crop manually. With hyperspectral imaging analysis in combination with appropriate analysis technique the images captured will help the agrarian to underrated and find the potential recognition of early stage of plant foliar disease and stress. The imaging used high-fidelity colour reflectance on the field that has wide range of light spectrum and identifies the subtle change in plant development and growth. There are various techniques used to determine plant stress and the data is collated by undertaking the following: quadratic discriminant analysis (wheat – yellow rust), multilayer perceptron (wheat yellow rust), and fishers’ linear determinant analysis etc. One technique is particular and has become popular for early detection of stress – simples volume maximization (SiVM). This technique helps in selecting spectral signatures that are sample of stressed and heathy plants – the data is clustered and the compared with pre-learned sample signatures. Other techniques to detect stress include support vector machine (SVM) – an algorithm labelled for healthy or drought plants and combined techniques include least squares support vector machine (LSSVM) and detect drought stress method. Further, these techniques are for machine learning and data experts who have experiences in pre-processing of data.