Classification of the Maturity of Tea Leaves using Hyperspectral Imaging

K. Subramanian; Kangkan Talukdar; Mamidi H Varshith; Krtyush Kumar

doi:10.63278/mme.v31i1.1246

Authors

K. Subramanian Assistant Professor, Department of IT and Analytics, Xavier Institute of Management and Entrepreneurship (XIME), India
Kangkan Talukdar Research Student, Department of IT and Analytics, Xavier Institute of Management and Entrepreneurship (XIME), India
Mamidi H Varshith Research Student, Department of IT and Analytics, Xavier Institute of Management and Entrepreneurship (XIME), India
Krtyush Kumar Research Student, Department of IT and Analytics, Xavier Institute of Management and Entrepreneurship (XIME), India

DOI:

https://doi.org/10.63278/mme.v31i1.1246

Keywords:

Hyperspectral Imaging, HIS in tealeaves.

Abstract

Technologies for the food industry to enhance product quality through precise and efficient analysis methods have been driven by advancements in the food industry, which have in turn led to the adoption of cutting edge technologies. HSI, which was first developed for remote sensing, has also been used for food commodity classification because it is a non-invasive technology that can analyze multiple chemical properties at a single time. Black tea production, the tender leaves (bud and first two leaves) must be harvested, but since manual plucking involves mature leaves, the quality of the product is reduced. A machine learning based classification model using hyperspectral imaging to identify tender tea leaves is presented in this research with the aim of improving productivity and minimizing plucking errors.

Spectral data of tea leaves were captured using a relatively inexpensive in-house hyperspectral camera using hyperspectral imaging. About six images, each containing 5–6 leaves, were used to get spectral information and were further divided into two maturity classes: tender and mature. The data was split into training and testing sets for the study. The classification performance of six algorithms: CART, LR, LDA, KNN, NB, and SVM was evaluated through confusion matrices.

The study demonstrated that among the six machine learning models tested, KNN and SVM achieved the highest accuracies of approximately 75%. When tested with the validation dataset, the models showed better performance in identifying tender leaves compared to mature leaves, which is particularly beneficial for tea manufacturers who primarily seek tender leaves. The research also successfully proved that a portable, low-cost hyperspectral imager could be constructed and calibrated for practical use, although software limitations prevented full utilization of the custom-built device. Despite using third-party hyperspectral tea leaf data for the final analysis, the study conclusively showed that tender leaves can be distinguished using HSI and machine learning, offering a promising solution for tea gardens to automate and improve their leaf selection process.

References

“Benelli, A., Cevoli, C., & Fabbri, A. (2020). In-field hyperspectral imaging: An overview on the ground-based applications in agriculture. Journal of Agricultural Engineering, 51(3), 129-139.”

“ElMasry, G., & Sun, D. W. (2010). Principles of hyperspectral imaging technology. In Hyperspectral imaging for food quality analysis and control (pp. 3-43). Academic Press.”

“Gowen, A. A., O'Donnell, C. P., Cullen, P. J., Downey, G., & Frias, J. M. (2007). Hyperspectral imaging–an emerging process analytical tool for food quality and safety control. Trends in Food Science & Technology, 18(12), 590-598.”

“Jia, Y., Huang, W., Guo, L., Wu, J., & Fan, S. (2018). Apple surface pesticide residue detection method based on hyperspectral imaging. In International Conference on Intelligent Science and Big Data Engineering (pp. 27-63). Springer, Cham.”

“Jian, W., Xianyin, Z., & ShiPing, D. (2010). Identification and grading of tea using computer vision. Applied Engineering in Agriculture, 26(4), 639-645.”

“Li, X., Wei, Y., Xu, J., Feng, X., Wu, F., Zhou, R., & Jin, J. (2017). Hyperspectral imaging and their applications in the nondestructive quality assessment of fruits and vegetables. In Hyperspectral imaging in agriculture, food and environment (pp. 27-63).”

“Salazar-Vazquez, J., & Mendez-Vazquez, A. (2020). A plug-and-play hyperspectral imaging sensor using low-cost equipment. HardwareX, 7, e00087.”