Application of odor imaging sensor coupled with hyperspectral imaging technology in monitoring the large yellow croaker (Larimichthys crocea) freshness

Total volatile base nitrogen (TVB-N) is an important indicator for evaluating the freshness of aquatic products and holds great significance in assessing food safety. Traditional testing methods for TVB-N content use the Kjeldahl method, which has shortcomings like lengthy processes, cumbersome steps, and sample destruction. This study innovatively couples the hyperspectral imaging (HSI) technique with an odor imaging sensor (OIS) to achieve non-destructive prediction of TVB-N content in the large yellow croaker. Gas changes in the fish were captured through OIS. After this, a hyperspectral imager is used to simultaneously characterize the hyperspectral information of the OIS and the fish sample itself. Subsequently, the fish eye and fish body regions were selected as two regions of interest (ROI) to represent the hyperspectral information of the fish sample. Comparing the modeling performance of HSI data from the ROI from the fish eye and body, it was found that the body ROI can better represent the corruption information of the large yellow croaker. Finally, data fusion was applied to analyze the optimization effect of paired fusion of spectral information from different data sources. It was found that the fusion of OIS and the HSI of large yellow croaker body parts can obtain an excellent partial least squares (PLS) model with a prediction set determination coefficient (R_p²) of 0.9506, providing a novel view for seafood safety assessment.