Packaged food detection method based on the generalized Gaussian model for line-scan Raman scattering images

Abstract

Packaged food safety has gained increasing attention worldwide. Existing analytical methods pose difficulties in accurately measuring food quality without destroying the packaging. In this study, a nondestructive detection method for packaged food was proposed based on the generalized Gaussian model for Raman scattering images. The Raman peaks of the scattering image were extracted, and the attenuation information of the peaks far from the laser point were imported into the established generalized Gaussian model. Analysis of the histogram of residual distribution revealed that the difference in residual distribution was enhanced, and an appropriate threshold was selected to separate the Raman baseline correction spectrum of the internal materials. Food-grade polyethylene sheets with thicknesses of 1, 2, and 3 mm were used as packaging materials for comparison experiments. The proposed model can accurately separate the Raman peak of the subsurface material when 1 mm-thick polyethylene was used as the packaging. Food-grade plastic sheets of polyethylene, polypropylene and high-density polyethylene were covered with pure substances such as melamine, sodium nitrite, and maleic anhydride. This model was considered suitable for most food-grade plastic packaging, and the subsurface materials did not influence the separation effect. Finally, evaluation of premium white granulated sugar demonstrated that the model effectively separated the Raman peak produced by packaged food and detected the packaged food without conferring damage.