Korea Food Research Institute (KFRI) said on the 16th that it developed a new technology that combines artificial intelligence (AI) and hyperspectral imaging to non-destructively evaluate quality changes that occur during the freezing and thawing of fish.
During the process of freezing fish for storage and then thawing it, various changes occur, including protein denaturation, moisture loss, and deterioration in texture. In particular, quality differences vary widely depending on the thawing method, such as room-temperature thawing or thawing under running water.
The KFRI research team compared room-temperature thawing and thawing under running water, the representative industrial thawing methods. After rapidly freezing mackerel at minus 20 degrees and thawing it using the two methods, they analyzed changes in components, microorganisms, texture, and color during storage.
At this time, the researchers used "near-infrared hyperspectral imaging," which can collect wavelength information from light reflected off the surface of mackerel. By analyzing a much wider range of wavelengths than a standard camera, it can detect subtle chemical differences invisible to the naked eye.
When near-infrared hyperspectral imaging was applied to an AI analysis model, differences between the two thawing methods were pronounced. On days 1–3 after thawing, room-temperature thawing and thawing under running water could be distinguished with an accuracy of about 95% or higher.
The researchers expect that if this technology is introduced to large-scale distribution and processing sites, it will be possible to build an automated quality control system and dramatically reduce inspection expense and time.
Park Seul-gi, a researcher at KFRI's Smart Manufacturing Research Group, said, "Hyperspectral imaging and AI analysis technology will provide a new paradigm for quality control and real-time analysis of seafood, including frozen seafood, as well as for quality and subspecies classification," adding, "It provides a foundation that could enable real-time, non-destructive analysis not only for the thawing process but also for other quality change analyses."