Non‐destructive beef adulteration detection using hyperspectral imaging and independent component analysis

Abstract BACKGROUND Meat adulteration poses significant food safety and economic fraud challenges, yet traditional detection methods are destructive, time‐consuming and unsuitable for rapid screening. Meanwhile, despite advances in non‐destructive and rapid hyperspectral imaging (HSI) technology, its predominant spectral‐biased approach faces challenges in distinguishing chemically similar adulterants. Notably, unlike authentic meat formed through natural processes, adulterated meat undergoes artificial reconstruction, creating distinct tissue surface morphological differences. These inherent differences offer key insights for rapid adulteration detection. Accordingly, this study developed a novel non‐destructive approach combining HSI with independent component analysis (ICA) to extract tissue surface features for beef adulteration detection. Two adulteration types were investigated: restructured minced meat and pork‐substituted beef at adulteration levels of 0%, 10%, 20%, 30%, 40% and 50%. RESULTS For both adulteration types, ICA was employed to extract tissue surface features from hyperspectral images, with derivatives characterizing surface‐change trends. Subsequently, k‐nearest neighbor and linear discriminant analysis (LDA) were applied to classify authentic and restructured minced beef, with the LDA model achieving perfect discrimination ( R 2 p = 1.0000, RMSEP = 0.0000). Meanwhile, for detecting pork‐substituted beef, hierarchical cluster analysis (HCA) and orthogonal partial least squares discriminant analysis (OPLS‐DA) were employed. The OPLS‐DA model significantly outperformed HCA and demonstrated excellent performance (AUC > 0.92 across all substitution ratios). Furthermore, the visualization successfully located pork‐substituted regions and quantified substitution levels. CONCLUSION This tissue surface feature‐based approach provides a rapid, accurate and non‐destructive method for meat adulteration detection, offering significant potential for industrial implementation, particularly suited for high‐throughput screening in meat processing facilities and quality assurance operations. © 2025 Society of Chemical Industry.