Tobacco Leaf NIR Management and Classification Based on JSON Conversion and Constrained K-Means for SQLite Database

The industry 4.0 is revolutionizing the way manufacturing processes in companies. The digitalization of material management and analysis requires flexible, convenient, and accurate software tools based on database technology. In the tobacco industry, the main material is tobacco leaves, which are frequently monitored using Near Infrared (NIR) Spectroscopy, generating large high dimensional NIR data sets. This paper presents a lightweight and portable C♯ software application built on SQLite database for managing and classifying tobacco leaves' NIR data. As the NIR spectra are of high dimension and variable length, some specialized techniques like JSON serializer and constrained K-means are used in this software.