The aim of the study was to detect samples of PSE meat (pale, soft and exudative) using the mass spectrometry method in the SWIR spectrum and highlight defects using chemical imaging.
The product to be analyzed were different samples of fresh meat. For these tests, a high-speed compact SWIR hyperspectral camera was used for industrial artificial vision applications. And for lighting, a halogen focus with linear concentration was used, a patented technology.
The tests were carried out on the conveyor belts installed in the INSPECTRA laboratory, placing the products on a plastic tray covered with absorbent paper.
The image processing software was used to capture samples through the hyperspectral camera. With this software, you can create an image in which the areas with different spectral shapes differ, that is, with different color tones depending on their chemical composition.
Here are the actual images of the samples compared to the chemical images obtained in the laboratory:
In Figure 1, it can be perfectly differentiated that the second sample from the left, is the only non-PSE.
Another case, comparing a PSE (left) with a non-PSE:
As can be seen in the chemical image, it is easily distinguished by this artificial vision software by comparing regions of different color tones, both parts with PSE (red color) and parts in good condition (green color), being able to set percentages of surface by number of pixels of different tones.
