This study introduces a novel two-step hierarchical framework for authenticating the geographical origin of freeze-dried instant coffee, combining digital image-based fingerprints and chemometrics. Color histograms in the Grayscale, RGB, and HSV color spaces served as sample fingerprints, which were used as input for constructing chemometric models. Using HSV histograms, Data-Driven Soft Independent Modeling of Class Analogy (DD-SIMCA) and Partial Least Squares Discriminant Analysis (PLSDA) demonstrated optimal predictive performance. These models accurately classified all samples from Southern Bahia and Espírito Santo in the first and second steps, respectively, and successfully differentiated them from samples of undeclared provenance. This resulted in 100 % sensitivity, specificity, and accuracy in the test set for both the class-modeling and discriminant approaches. The proposed methodology is simple, cost-effective, and non-destructive, offering an efficient, green, and reliable tool for in situ verification of instant coffee's authenticity and geographical origin.