Skip to main content
ARS Home » Northeast Area » Beltsville, Maryland (BARC) » Beltsville Agricultural Research Center » Environmental Microbial & Food Safety Laboratory » Research » Publications at this Location » Publication #386084

Research Project: Sensing Technologies for the Detection and Characterization of Microbial, Chemical, and Biological Contaminants in Foods

Location: Environmental Microbial & Food Safety Laboratory

Title: Simulated annealing-based hyperspectral data optimization for fish species classification: Can the number of measured wavelengths be reduced?

Author
item CHAUVIN, JOHN - University Of North Dakota
item DURAN, RAY - University Of North Dakota
item TAVAKOLIAN, KOUHYAR - University Of North Dakota
item AKHBARDEH, ALIREZA - University Of North Dakota
item MACKINNON, NICHOLAS - Collaborator
item Qin, Jianwei - Tony Qin
item Chan, Diane
item HWANG, CHANSONG - Us Forest Service (FS)
item BAEK, INSUCK - Orise Fellow
item Kim, Moon
item ISAACS, RACHEL - Chapman University
item YILMAZ, AYSE - Chapman University
item ROUNGCHUN, JIALEEN - Chapman University
item HELLBERG, ROSALEE - Chapman University
item VASEFI, FARTASH - Collaborator

Submitted to: Food Control
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 11/8/2021
Publication Date: 11/11/2021
Citation: Chauvin, J., Duran, R., Tavakolian, K., Akhbardeh, A., Mackinnon, N., Qin, J., Chan, D.E., Hwang, C., Baek, I., Kim, M.S., Isaacs, R., Yilmaz, A., Roungchun, J., Hellberg, R., Vasefi, F. 2021. Simulated annealing-based hyperspectral data optimization for fish species classification: Can the number of measured wavelengths be reduced?. Food Control. 11:10628. https://doi.org/10.3390/app112210628.
DOI: https://doi.org/10.3390/app112210628

Interpretive Summary: Many fish fillets are similar in appearance, which makes them a susceptible target for economically-motivated fraud. Mixing less expensive species into more expensive species is one of major fish fraudulent practices in the seafood industry. This study developed a data analysis methodology to support design of a future spectroscopy-based system for detecting mislabeling of fish fillets. Three types of spectra, including fluorescence and reflectance in visible and near-infrared and short-wave infrared regions were obtained from hyperspectral images acquired from the fish fillet samples of 25 common species. Algorithms were developed for wavelength selection, data fusion, and machine learning classification. Based on a multi-layer perceptron neural network classifier, a 95% classification accuracy was achieved using the fusion of the three spectral modes with seven wavelengths selected by a simulated annealing method. The data analysis methods developed in this study can facilitate development of a rapid and cost-effective spectral sensing device for on-site inspection of the fish fillet mislabeling, which can be used for authentication of the fish fillets and other related food products by the seafood industry and regulatory agencies.

Technical Abstract: Relative to standard red/green/blue (RGB) imaging systems, hyperspectral imaging systems offer superior capabilities but tend to be expensive and complex, requiring either a mechanically complex push-broom line scanning method, a tunable filter, or a large set of light emitting diodes (LEDs) to collect images in multiple wavelengths. This paper proposes a new methodology to support the design of a hypothesized system that uses three imaging modes—fluorescence, visible/near-infrared (VNIR) reflectance, and shortwave infrared (SWIR) reflectance—to capture narrow-band spectral data at only three to seven narrow wavelengths. Simulated annealing is applied to identify the optimal wavelengths for sparse spectral measurement with a cost function based on the accuracy provided by a weighted k-nearest neighbors (WKNN) classifier, a common and relatively robust machine learning classifier. Two separate classification approaches are presented, the first using a multi-layer perceptron (MLP) artificial neural network trained on sparse data from the three individual spectra and the second using a fusion of the data from all three spectra. The results are compared with those from four alternative classifiers based on common machine learning algorithms. To validate the proposed methodology, reflectance and fluorescence spectra in these three spectroscopic modes were collected from fish fillets and used to classify the fillets by species. Accuracies determined from the two classification approaches are compared with benchmark values derived by training the classifiers with the full resolution spectral data. The results of the single-layer classification study show accuracies ranging from ~68% for SWIR reflectance to ~90% for fluorescence with just seven wavelengths. The results of the fusion classification study show accuracies of about 95% with seven wavelengths and more than 90% even with just three wavelengths. Reducing the number of required wavelengths facilitates the creation of rapid and cost-effective spectral imaging systems that can be used for widespread analysis in food monitoring/food fraud, agricultural, and biomedical applications.