Weak-Signal Spectral Detection in Meat Quality Monitoring

Abstract

Meat quality monitoring requires sensitive analytical methods capable of detecting contamination signals before they become visually apparent or chemically dominant. Hyperspectral imaging offers high-dimensional spectral information that can capture subtle biochemical and textural changes in meat products. This topic focuses on weak-signal spectral detection, where contamination signatures are present but mixed with dominant tissue, fat, moisture, and illumination effects. Hyperspectral unmixing benchmarks provide structured evaluation settings for comparing algorithms under realistic contamination levels. Spectral preprocessing, dimensionality reduction, endmember learning, and abundance estimation are central to improving detection reliability. Food safety applications also require interpretable models, since inspection decisions must be linked to understandable spectral evidence rather than opaque classification outcomes. This literature direction supports automated meat quality monitoring systems that combine spectral sensitivity, robust benchmarking, and practical food safety relevance.