Report Detail Page
Hyperspectral Food Safety Inspection System
With the current growing need for low production costs and high efficiency, the food industry is faced with a number of challenges, including maintenance of high quality standards and assurance of food safety while avoiding liability issues. A hyperspectral camera with a large spectral range (400-2500nm) was purchased to investigate its ability to detect contaminations of interest linking to food safety issues such as ingesta, faeces, bile, urine, Salmonella, E. coli and Listeria. While some success was achieved with all contaminants, ingesta and faeces were the most successful. These were also deemed to be of high value to industry as importing countries use them as hygiene indicators. The second set of trials conducted focussed on ingesta and faeces. Detection algorithms were developed using two different analysis methodologies, both yielding high levels of accuracy - up to 99.67% success rate for classification of pixels as contamination, fat or lean pixels in the hyperspectral image. The algorithms were built focussing on a number of key wavelengths, particularly within the 450-750nm and 900-1450nm ranges.
While challenges still exist to transition to commercial scale, it is believed that the hyperspectral imaging technology is suitable for performing on-line detection of faeces and ingesta contamination on red meat products.
It is recommended that the results and algorithms developed in this project should be built upon by conducting further trials examining a larger sample size to improve accuracy, demonstrate robustness in coping with variations in carcase factors (e.g. area of the carcase, breed, feed type) and help develop a commercially viable concept system.
This page was last updated on 25/07/2017
Join myMLA today
One username and password for key integrity and information Systems (LPA/NVD, NLIS, MSA & LDL).
A personalised online dashboard that provides news, weather, events and R&D tools relevant to you.
Customised market information and analysis.
Already registered for myMLA?