An advanced instrument for UV-Vis spectra digitization: A progression powards establishing a saffron metabolite database for AI-driven quality assurance

The verification of authenticity and quality of saffron, the most precious spice globally, has become a challenging task, especially for small-scale producers that necessitate dependable on-site monitoring to avert fraud. A significant obstacle to the advancement of contemporary, data-driven solutions is that decades of essential UV-Vis spectral data for saffron's primary metabolites are confined as static numbers in published literature, rendering them inaccessible for AI modeling. To address this issue, we present an AI-assisted software tool specifically designed for the extraction, digitization, and enhancement of these old spectra. The software combines an intuitive graphical interface with a sophisticated image processing pipeline, incorporating automated curve skeletonization and accurate k-d tree point tracing to guarantee precise data acquisition from low-quality photos. A thorough data refinement process enables users to implement advanced smoothing, diverse interpolation techniques, baseline correction, and interactive point editing to recover pristine, high-fidelity spectral curves. We effectively employed this method to digitize and enhance several saffron metabolite spectra from various publications, resulting in a curated, machine-readable database. This fundamental dataset is crucial for training sophisticated AI models aimed at quality control and anti-adulteration, establishing the technological basis for automated fraud detection and real-time quality assurance in the saffron supply chain.

کلیدواژه ها:

Data digitization ، Quality control ، Saffron authentication ، Image processing