Photonics and the Future of Manufacturing and Agriculture

By Rebecca Ingle, Ph.DFeb 21 2022

Photonics technologies are already revolutionizing many industries. A key part of this is developments in sensing technologies and image processing algorithms. The combination and the power of the two have led to sufficient demand and photonics has become one of the UK’s most productive manufacturing sectors.¹

Image Credit: Fotokostic/Shutterstock.com

The agricultural industry is one example where photonics has helped improve process efficiency and reduce waste. Traditional farming methods rely extensively on time-consuming processes that must be carried out manually.² This means a large labor force is required which adds significant expense to food growth and processing.

Photonics is helping automate processes such as screening fields for crop quality and plant diseases. With the right spectroscopic and imaging techniques, it is possible to check water levels in plants and determine the nutritional content of plant species.

The data that can be collected from plants in the field is used to refine agricultural practices in a more precise way – it is possible to monitor exactly how plant growth and the final product composition changes in response to farming practices.

Being able to collect real-time data to establish best practices and help maximize yields is more crucial than ever due to the challenges farmers face trying to cope with pesticide resistance, minimizing pesticide use, and crop growth issues with changing climate conditions.

Machine Vision

One of the biggest applications in photonics for agriculture is machine vision. ‘Agricultural robots’ or drones for crop inspection can be used for ripeness inspection of size profiling of produce.³ The advantage of using drones for this is that many field areas span several kilometers, with improvements in autonomous navigation meaning that robots can be left to independently survey such areas.

For brightly colored fruits such as apples, grapes and flowers, convolutional neural networks are often used to perform color sorting of the captured images to discriminate against the background of plant foliage and the environment. Some models also make use of fruit shapes to identify parameters such as ripeness.

Visible light imaging is not the only option for agricultural monitoring. Particularly for soils and screening for plant diseases, infrared imaging has become a popular choice.⁴ Depending on the exact spectral range, infrared imaging can be used to profile the thermal properties of plants and soils or as a way of profiling the chemical composition of minerals and other species. Using spectroscopy and imaging applications for profiling soils was previously carried out using offsite analysis. With developments in the automation of the sampling and analysis, this can now be done on-site.

There are limitations to the power of machine vision, particularly with visible light vision applications, including:

• Leaves obscuring produce, making it difficult to capture sufficiently high signal-to-noise images
• Overlapping fruit can cause issues with fruit counting³

However, the development of new software and sorting algorithms will help make autonomous agricultural robots more efficient.

Advances in robotics are also making autonomous fruit picking vehicles a possibility.⁵ Fruit picking represents a particularly challenging task for robots as it requires precise recognition of where the produce is located and, as many types of produce are also very soft and easily damaged in the picking process, very careful application of force avoids this.

Hyperspectral Imaging

A more advanced application of machine vision is hyperspectral imaging. In hyperspectral imaging, a full spectrum is recorded at each point of the image. This results in a dataset where molecular level information can be recovered with spatial resolution from each region of the image.

Depending on the wavelength region used for hyperspectral imaging, different information can be recovered on the molecular species present. These information-rich datasets can be used to examine plant properties such as chlorophyll levels and calculate plant biomass, estimate nitrogen concentrations, and are particularly valuable when looking for markers of plant disease.⁶

While a standard imaging camera can only look for the same visible markers of disease that a human would be sensitive to such as wilted leaves or discoloration, hyperspectral imaging is sensitive to the molecular level changes that occur on infection – often days before a plant shows any visible symptoms. This allows for much earlier intervention and more effective treatments.

Another application of hyperspectral imaging is for crop type identification. By using the typical chemical profiles of plants as well as visual markers for different plant types, hyperspectral imaging can be used to accurately discriminate against different plant species. This can also be extended to the identification of invasive species. While it can be challenging to differentiate between crops and plants on leaf color alone, hyperspectral imaging can efficiently differentiate between crops and unwanted species. This information can then be used for manual or automated removal of the weeds.

Looking Ahead

Photonics is affecting all aspects of agricultural processes, from the planting and growth of crops to product manufacturing. Infrared spectroscopy has been a staple of the dairy industry for checking milk quality and is now increasingly used as an online analysis tool for dairy product processing.⁷

As image recognition algorithms continue to advance, it may be that robots become a commonplace site on farms.

References and Further Reading 

1. Photonics Leadership Group (2022) Photonics by 2035, https://photonicsuk.org/wp-content/uploads/2021/10/Photonics_2035_Vision_Web_1.0.pdf, accessed February 2022
2. Yeong, T. J., Jern, K. P., Yao, L. K., Hannan, M. A., & Hoon, S. T. G. (2019). Applications of photonics in agriculture sector: A review. Molecules, 24(10), 1–39. https://doi.org/10.3390/molecules24102025
3. Mavridou, E., Vrochidou, E., Papakostas, G. A., Pachidis, T., & Kaburlasos, V. G. (2019). Machine Vision Systems in Precision Agriculture for Crop Farming. Journal of Imaging, 5, 89. https://doi.org/10.3390/jimaging5120089
4. Aliah, B. S. N., Kodaira, M., & Shibusawa, S. (2013). Potential of visible-near infrared spectroscopy for mapping of multiple soil properties using real-time soil sensor. In Sensing Technologies for Biomaterial, Food, and Agriculture 2013, International Society for Optics and Photonics, 8881, 888107.
5. Chiu, Y. C., Chen, S., & Lin, J. F. (2013). Study of an autonomous fruit picking robot system in greenhouses. Engineering in agriculture, environment and food, 6(3), 92-98. https://doi.org/10.1016/S1881-8366(13)80017-1
6. Mavridou, E., Vrochidou, E., Papakostas, G. A., Pachidis, T., & Kaburlasos, V. G. (2019). Machine Vision Systems in Precision Agriculture for Crop Farming. Journal of Imaging, 5, 89. https://doi.org/10.3390/jimaging5120089
7. De Marchi, M., Toffanin, V., Cassandro, M., & Penasa, M. (2014). Invited review: Mid-infrared spectroscopy as phenotyping tool for milk traits. Journal of Dairy Science, 97(3), 1171-1186. https://doi.org/10.3168/jds.2013-6799

Disclaimer: The views expressed here are those of the author expressed in their private capacity and do not necessarily represent the views of AZoM.com Limited T/A AZoNetwork the owner and operator of this website. This disclaimer forms part of the Terms and conditions of use of this website.