Juhi Ranjan, Rangoli Mahesh, Shivam Chaubey PhD Research Scholar,
ICAR-IARI, New Delhi
Ramesh Sahni, Scientist, ICAR-CIAE, Bhopal

Agricultural materials exhibit diverse chemical compositions and internal physical structures. Consequently, when utilizing near-infrared (NIR) spectroscopy, these materials interact with electromagnetic energy at specific wavelengths in distinct ways, through reflection, scattering, absorption, and emission. Near-infrared (NIR) spectroscopy is a non-destructive examination tool that has long been utilized in agriculture. It is used to quickly test the nutritional content of dried and fresh food materials, as well as to assure food safety. NIR technology has recently been merged with microscopy to form NIR microscopy, and with imaging methods to form hyperspectral imaging (HSI). HSI is a combination of visible/near-infrared spectroscopy and vision methods that has the ability to obtain more accurate and detailed information than typical NIR technologies. NIR-HSI goes beyond traditional spectroscopy by integrating spatial location information from the collected spectra.Despite the reduction in wavelength resolution, the NIR-HSI spectrum compensates by providing better spectral quality generated from hundreds of spectra. A NIR spectroscopy device produces a single spectrum every measurement, whereas hyperspectral pictures produce hundreds of spectra from a single sample. Each pixel corresponds to one spectrum in each measurement. Furthermore, the NIR-HSI picture captures the sample's distinctive spectral signature, allowing for the characterisation and identification of various materials.

NIR spectroscopy and HSI are significant agricultural technologies with the potential to alter the way we produce and manage food. It is a non-destructive analysis approach based on sample diffuse reflectance. This technique is widely used for determining the concentration of nutrients and feed value in both dried and fresh crop materials, as well as for quality control of food and feed and assuring food safety.Hyperspectral pictures were initially employed in remote sensing applications such as detection and mapping. Because of their broad spectral range and great spatial resolution, these pictures are perfect for mapping and understanding the Earth's surface. They may also be used to identify soil qualities including moisture, organic matter concentration, and salinity. Hyperspectral pictures, in addition to remote sensing, have been employed in the paper sector. NIR-HSI can sort a variety of materials, including pulp, paper, cardboard, newspaper, bleached and unbleached fibers. This technique has the potential to increase the efficiency and quality of paper manufacturing. The authors of the mentioned studies (Ben-Dor et al., 2002; Tatzer et al., 2005) go into further depth regarding the usage of hyperspectral pictures in remote sensing and the paper industry.

Basics of hyperspectral imaging (HSI) technology
A good knowledge of the core concepts of HSI is required to use this powerful approach successfully. In most cases, the process begins with hardware that takes pictures and software that processes them. A light source, detector, wavelength dispersion device, and a computer with image collection software are all required components of any HSI system. It is critical to carefully choose these components in order to get optimal performance and dependable, high-quality hyperspectral pictures.There are two types of light sources in HSI systems: illumination sources and excitation sources. Broadband lights are often used in illumination sources, but narrowband lights are commonly used in excitation sources (Qin et al., 2013).Tungsten halogen lamps (THL) are popular due to their long life, reliability, and ability to provide a smooth spectrum from visible to infrared wavelengths. THLs are also affordable (Wu and Sun, 2013). THLs, on the other hand, release a substantial quantity of heat, which might possibly affect the physical and chemical characteristics of the samples. Broadband light-emitting diodes (LEDs) are also becoming more common in HSI systems because to its benefits such as extended lifespan, low power consumption, minimal heat production, compact size, rapid reaction, durability, and vibration resistance (Qin et al., 2013). Lasers, on the other hand, are strongmonochromatic sources that are widely utilized for excitation.They are suitable for fluorescence and Raman studies due to their concentrated energy, accurate directionality, and monochromatic emissi=on (Qin et al., 2013).

The detector's performance has a direct impact on image quality. A detector with high sensitivity and a high signal-to-noise ratio is often required. Charge-coupled devices (CCDs) and complementary metal-oxide semiconductors (CMOS), two extensively used detectors, have advanced rapidly. Both types of detectors are sensitive to visible and short wavelength near-infrared areas (400-1000 nm), however their efficiency reduces dramatically below 400 nm and increases greatly beyond 900 nm. More costly detectors, such as Indium Gallium Arsenide (InGaAs) or Mercury Cadmium Telluride (MCT), are frequently employed for near-infrared imaging (900-2500 nm).Wavelength dispersion devices are crucial in HSI. These devices are used to choose the excitation wavelength between the light source and the sample, or to segregate the emitted wavelengths between the sample and the detector. Their primary function is to scatter light into different wavelengths and direct the dispersed light to the area detector. This may be accomplished with a variety of optical and electro-optical instruments, such as image spectrographs, tunable filters such as acousto-optic tunable filters (AOTFs) and liquid crystal tunable filters (LCTFs), and beam-splitting devices. In the development of line-scan HSI systems for applications such as food quality and safety inspection, imaging spectrographs have been frequently employed.

Because of its electronic control, absence of moving components, and quick tuning rates, AOTFs and LCTFs are increasingly being used in area-scan HSI systems for agricultural applications. To assist spectrum imaging approaches, all HSI systems now on the market are integrated and comprise important components such as a light source, wavelength dispersion device, detector, and computer with image capture software. However, researchers can build their own system by merging all of the main components.The image capture technique (whisk broom, gazing, or push broom), spectral ranges (UV, visible, near-infrared, mid-infrared), and measuring mode (reflectance, transmittance, or interactance) of HSI systems are all different.The acquisition mode, which governs how spectral and spatial information is gathered, is the basic categorization strategy for HSI systems. The traditional HSI system employs two scanning methods: spatial (point and line scanning) and spectral (area scanning). As a result, there are three methods for creating a hypercube: point scanning, line scanning, and area scanning. A spectrum is collected at a single spatial position in point scanning (whisk-broom imaging), and subsequently further locations are scanned by moving either the detector or the sample. This procedure must be performed for each geographic point that requires spectral data, making it the most time-consuming technique of acquiring HSI data.

Line scanning (push-broom imaging) gathers spectral information line by line and necessitates relative movement between the sample and the detector. It is 100 times quicker than point scanning and is ideal for online quality control. The hypercube is acquired sequentially, one wavelength at a time, using area scanning (staring imaging). A dispersion device (spectrograph) is not required in this procedure because the entire scene is photographed at multiple wavelengths. Area scanning does not need any relative movement between the detector and the item. The area scanning technique, on the other hand, is only usable with multispectral or comparable imaging systems with a restricted number of wavelengths.Spectral scanning techniques often store the hypercube in a band-sequential format, which compromises the balance of spatial and spectral information, whereas spatial scanning saves the hypercube in either the form or the form.

Hyperspectral applications in agriculture
In agriculture, HSI has found diverse applications encompassing various objectives. These include estimating biochemical properties of crops such as chlorophyll, carotenoids, and water contents, as well as biophysical properties like Leaf Area Index (LAI) and biomass. These measurements aid in comprehending the physiological state of vegetation, predicting crop yield, assessing nutrient status (such as nitrogen deficiency) in crops, monitoring diseasesin crops, and examining soil properties such as soil organic matter, soil moisture, and soil carbon.

1. Crop biochemical and biophysical properties estimation
HSI has found a significant application in agriculture by enabling the monitoring of crop conditions through the extraction of crop biochemical and biophysical properties. One crucial property is the leaf chlorophyll content, which plays a vital role in the photosynthetic capacity of vegetation and overall crop productivity.In addition to leaf chlorophyll, the water content of crops serves as a vital parameter for indicating water stress. Previous studies have primarily focused on estimating crop chlorophyll content, Leaf Area Index (LAI), and water content using HSI. However, other important crop properties, such as carotenoids, which serve as sensitive indicators of plant stress, have not been explored as extensively. It is crucial to not only investigate the spatial and temporal variations of each property but also assess the relationships between these properties to gain a deeper understanding of their impacts on crop growth and production.Estimating crop biomass and predicting yield are also significant applications of remote sensing in agriculture, as they contribute to the understanding of crop productivity and facilitate appropriate management practices. Additionally, the crop residues left in fields play a critical role in protecting the soil from erosion caused by water and wind, while also influencing soil biochemical processes. Several studies have utilized hyperspectral images to estimate crop residues in agricultural lands. Moreover, an emerging research topic involves exploring the potential of bioenergy generation, such as biogas, from crop biomass.

In summary, hyperspectral imagery has made significant contributions to the estimation of crop biomass, yield, and related aspects such as bioenergy and crop residues. Since agricultural practices, such as watering and nutrient treatments, have a substantial impact on crop biomass and yield, incorporating data on these practices along with hyperspectral imagery into models can potentially lead to improved results.

2. Crop nutrient status evaluation
Precision farming is an agricultural practice that involves evaluating the nutritional needs of crops and providing targeted management recommendations specific to each crop's requirements. This approach is essential for maximizing resource efficiency and minimizing negative environmental effects. HSI provides a wealth of spectral data, which enables precise assessment of crop nutrient levels. This, in turn, allows for the development of accurate fertilizer treatment plans aimed at achieving optimal crop yields. However, it is crucial to acknowledge that several factors, such as soil moisture, soil type, and topography, exert significant influence on crop growth and productivity. To have a more substantial impact on crop production, it is advisable to develop comprehensive treatment plans that take into account not only the nutritional status of the crops but also these other influential factors. By considering the broader context of the crop's environment, farmers can make informed decisions that promote sustainable and efficient agricultural practices.

3. Classifying imagery to identify crops, growth stages, weeds and stresses
In addition to measuring crop properties, hyperspectral images have proven to be useful in various classification tasks related to agriculture. These tasks include distinguishing between different types of crops, identifying different stages of crop growth, classifying weeds or invasive species, and detecting diseases. Agricultural land covers and crop types have distinct spectral characteristics, making hyperspectral images highly valuable for accurately classifying these agricultural features. Weed infestation is a significant challenge in agricultural fields and can have a profound impact on crop growth and yield. Remote sensing techniques that utilize HSI can aid in the identification and mapping of weeds in agricultural fields, which would greatly facilitate the implementation of variable rate treatments. It's important to note that the identification of weeds often requires high spatial resolution since many weeds are small and intertwined with crops. Close-range HSI conducted by Unmanned Aerial Vehicles can capture high-resolution images, offering significant potential for effective weed detection.

Monitoring crop diseases is crucial for growers who aim to mitigate economic losses and yield reductions. HSI allows the collection of signals at fine spectral resolutions, often with intervals of less than 10 nanometers. As a result, it has the potential to detect early symptoms of crop diseases and enable timely interventions. Hyperspectral signals are sensitive to variations in crop growth caused by diseases or stress, making them effective indicators of crop disease or stress occurrence. However, it's important to consider that crop status can be influenced by other factors, such as nutrient deficiency. Therefore, conducting repeated imaging and analysis, along with robust modeling techniques, is critical to ensure accurate and timely detection of crop diseases or stress.

4. Estimating soil physical or chemical properties
Soil properties in agriculture, including soil moisture, soil organic matter, soil salinity, and roughness, have a significant impact on crop growth and overall agricultural production. Hyperspectral remote sensing presents a promising approach for studying and analyzing these factors. Although optical remote sensing data can be utilized to estimate soil moisture, its accuracy is often affected by the presence of vegetation covering the ground. To improve the accuracy of soil moisture estimation, it is beneficial to integrate multiple types of remote sensing data, such as Synthetic Aperture Radar (SAR) and thermal data.

Soil organic carbon (SOC) is a critical component of soil fertility, exerting a substantial influence on crop growth and yield. Hyperspectral data, with its detailed spectral information, plays a vital role in accurately estimating the content of SOC. The use of hyperspectral imagery holds significant potential for estimating soil organic matter and carbon. However, similar to the challenges faced in soil moisture estimation, the presence of vegetation cover can significantly affect the estimation of soil organic matter and carbon. One potential solution to mitigate this issue is to collect hyperspectral images during non-growing seasons when vegetation cover is minimal.

In addition to soil moisture and organic matter, hyperspectral remote sensing data can be employed to estimate various other soil characteristics. Different soil features have distinct effects on spectral signals across different bands, and these influences can vary in magnitude. Furthermore, some of these influences may overlap spectrally. Therefore, when investigating a specific soil feature, it is crucial to collect an appropriate number of soil samples while considering the presence of other soil features that typically exert control over the spectral signals. This comprehensive approach ensures accurate characterization of individual soil characteristics while accounting for potential spectral interferences from other soil properties.

Advantages and disadvantages of HSI
HSI offers numerous advantages, including easy data acquisition, cost-effectiveness per analysis, quick inspection, simultaneous analysis of multiple compounds, nondestructive nature, and high accuracy. However, in near-infrared (NIR) spectroscopy systems, samples usually need to be ground to less than 1 mm. In contrast, NIR-HSI systems eliminate the need for sample preparation, such as grinding, allowing the samples to be utilized for other purposes. This not only saves time in sample preparation but also in database registration. NIR-HSI can capture thousands of spectra, providing a comprehensive view of the distribution of chemical compounds at the pixel level and enabling the simultaneous capture of spectral and spatial information. Hyperspectral images excel in providing high-quality surface spectra while also revealing internal details, such as detecting and quantifying the distribution of bacteria within a product.

Despite its potential for disease and defect detection in agricultural products and food, the application of NIR-HSI is limited by the costs of equipment. Additionally, achieving rapid image acquisition and analysis with NIR-HSI requires highly efficient hardware speed. Since NIR-HSI is an indirect method, calibration models are necessary. Combining NIR-HSI with chemometric techniques becomes essential to achieve effective qualitative and quantitative analyses. Chemometrics involves the use of mathematical and statistical methods to extract and interpret chemical information from data. However, a drawback is that building and processing such models can be time-consuming. Moreover, interpretation programs are often expensive, and expertise is required for calibration and standardization. Another challenge of NIR-HSI is the registration of successive overlapping bands, which makes it difficult to assign them to specific chemical groups and work with what are perceived as poor-quality pixels. To identify and detect unambiguous spectra within the same image, the samples must exhibit the same absorption characteristics.

HSI holds great promise for the future of agriculture, especially in precision agriculture, as it has the capability to capture highly detailed spectral information that is sensitive to various biophysical and biochemical properties of plants and soil. In recent times, there has been a notable increase in the availability of multiple platforms like satellites, airplanes, UAVs (unmanned aerial vehicles), and close-range platforms, which enable the collection of hyperspectral images with diverse spatial, temporal, and spectral resolutions. The application of HSI in agriculture has demonstrated successful outcomes across a broad range of areas. These applications encompass the estimation of crop biochemical and biophysical properties, evaluation of crop nutrient and stress levels, classification and detection of different crop types, weeds, and diseases, as well as the examination of soil characteristics.

While previous research has predominantly focused on integrating one or two factors that influence crop growth and productivity, it is crucial to take into account the integration of multiple factors in order to comprehensively evaluate crop status and identify growth-limiting factors. By incorporating these various factors, a more comprehensive understanding of their interconnectedness can be achieved, ultimately leading to optimized crop production and environmental protection.


1. Ben-Dor, E., Patin, K., Banin, A., and Karnieli, A. (2002) Mapping of several soil properties using DAIS-7915 hyperspectral scanner data. A case study over clayey soils in Israel. Int. J. Rem. Sens., 23 (6): 1043-1062.

2. Dai, Q., Cheng, J.-H., Sun, D.-W., Zeng, X.-A., 2015b. Potential of hyperspectral imaging for non-invasive determination of mechanical properties of prawn (Metapenaeusensis). Journal of Food Engineering 136, 64-72.

3. Garini, Y., Young, I.T., McNamara, G., 2006. Spectral imaging: principles and applications. Cytometry Part A 69A, 735-747.

4. Tatzer, P., Wolf, M., and Panner, T. (2005) Industrial application for inline material sorting using hyperspectral imaging in the NIR range. R. Time Imag., 11: 99-107.

5. Qin, J., Chao, K., Kim, M.S., Lu, R., Burks, T.F., 2013. Hyperspectral and multispectral imaging for evaluating food safety and quality: a review. Journal of Food Engineering 118, 157-171.

6. Wu, D., Sun, D.-W., 2013a. Application of visible and near infrared hyperspectral imaging for non-invasively measuring distribution of water-holding capacity in salmon flesh. Talanta 116, 266-276.