Fingerprints based on digital images integrated with DD-SIMCA for detection of food fraud in type 1 rice in the Brazilian market

A chemometrics-assisted color histogram-based analytical system (CACHAS) was evaluated as a simple, low-cost and non-destructive approach for the authentication of Type 1 rice adulterated with broken grains and rice fragments. Fingerprints were obtained from grayscale, RGB and HSI color histograms and used as analytical descriptors. Data-driven soft independent modeling of class analogy (DD-SIMCA) was applied to the extracted color histogram fingerprints in order to authenticate Type 1 rice samples and detect adulteration with broken grains and rice fragments. Models were developed at significance levels of α = 0.01 and α = 0.05, with the number of principal components optimized using the compliant one-class classification approach. The model integrating grayscale + RGB + HSI information showed the best performance, achieving 100% sensitivity, 100% specificity and 100% efficiency in the prediction stage, with stable behavior in the external test set, and a specificity of 96.4%. The observed discrimination is associated with morphological and structural differences between whole and broken/fragmented rice grains, which modify the interaction of light and diffuse scattering patterns, producing measurable variations in color histogram responses. The proposed approach demonstrates strong potential for the authenticity assessment and quality control of Type 1 rice sold in the Brazilian market. • Adulterations with byproducts from the rice processing process. • Morphocolorimetric profiles were obtained from color histograms of digital images. • Data-based class modeling was used to authenticate Brazilian Type 1 rice. • Adulterations with rice fragments and broken grains of up to 1% were detected. • The grayscale+RGB + HSI model authenticated Type 1 rice with 99.4% efficiency.