On-site saffron origin identification using image processing and chemometric tools

Abstract

Saffron (Crocus sativus L.), commonly known as ‘Red Gold’ is highly prized for its medicinal properties but is labor-intensive, requiring meticulous hand-harvesting, and is vulnerable to adulteration. Additionally, its price varies considerably depending on its origin, underscoring the need for robust surveillance to ensure authenticity. However, traditional methods for origin authentication are often complex and costly, posing challenges, especially for small cooperatives that have a major role in saffron distribution. This study presents an innovative approach to saffron provenance using digital imaging as an alternative on-site method. 118 saffron samples from Morocco (Taroudant, Ouarzazate, and Azilal), Afghanistan, Iran, Spain, and Tunisia were analyzed. Digital images were taken with a smartphone, and various color spaces were evaluated by the open-source software ImageJ, including RGB (Red-Green-Blue), HSB (Hue-Saturation-Brightness), LAB (Lightness-Green to Red-Blue to Yellow), and YUV (Luminance and Chrominance components), resulting in 2,712 variables per saffron sample. The collected data were then analyzed by chemometric tools. Principal component analysis showed strong separation and sample grouping, enabling effective screening of saffron origin based on the calculated image parameters, with the first three principal components explaining a significant variance (70–92%). Hierarchical clustering analysis also demonstrated clear clustering for most samples, while linear discriminant analysis achieved high classification accuracy (around 96%). Furthermore, partial least squares provided excellent calibration results for predicting saffron pigmentation based on image-derived data with an R² of 0.998 and RMSEC between 0.136 and 0.213.