Scientific journal
Journal of Food and Nutrition Research
Summary No. 2 / 2020
Mladenov, M. I.
Model-based approach for assessment of freshness and safety of meat and dairy products using a simple method for hyperspectral analysis
Journal of Food and Nutrition Research, 59, 2020, No. 2, s. 108-119
Mirolyub Ivanov Mladenov, Department of Automatics and Mechatronics, Faculty of Electrical Engineering, Electronics and Automation, University of Ruse “A. Kanchev”,
Received 29 November 2019; 1st revised 1 March 2020; 2nd revised 3 April 2020; accepted 22 April 2020; published online 14 May 2020.
Summary: A model-based approach for assessment of freshness and safety of meat and dairy products is proposed. In contrast to traditional methods for assessment of food quality, this approach can be used for rapid and non-destructive evaluation „on site“, i.e. in food shops, warehouses, catering services or homes. Freshness and safety are evaluated using three types of regression predictive models. A new simple method for hyperspectral analysis is proposed. Extraction and selection of the food quality features are supported by this method. It avoids the basic limitations of the hyperspectral imaging technologies for evaluating food quality and safety. It reduces sufficiently the number of bands of the hyperspectral characteristics in comparison with the traditional methods used in hyperspectral imaging. The method uses a new criterion for bad selection. The complex optical system for light dispersion in the hyperspectral imaging sensor could be replaced with an optical system of a traditional spectrophotometer by using this method. The food quality features are extracted from the spectral characteristics of the spectrophotometer and selected frequency bands of the hyperspectral characteristics. The separation error between different spectral data classes is evaluated. It sufficiently decreases when the spectral data from a selected frequency band are used.
Keywords: model-based approach; hyperspectral analysis; food safety; predictive model
Download:
(pdf, 826.7 Kb, 2065x)