Detection of bone fragments and other physical contaminations in deboned poultry meat has become increasingly important to ensure food quality and safety. Traditional X-ray imaging detection technologies have significant difficulties detecting contaminations because of the meat tissue thickness variation.

In order to address the thickness variation problem, in this study, a novel vision system with combined X-ray and laser 3D imaging technology has been developed for accurate physical contamination detection. The X-ray part of the combined system captures high resolution X-ray images in real-time, and the laser 3D part provides an accurate thickness profile for each piece of meat. In the combined system, the 3D thickness information is used to cancel the thickness variation in the X-ray image, thus the process of physical contamination detection is significantly simplified.

The combined vision system is capable of detecting calcified bones (rib bones and pulley bones) at a 95% detection rate, and partially calcified bones (fan bones) at a 90% detection rate.

In order to handle the inspection tasks in real-time, a multithread architecture is used in this vision system. Various threads work simultaneously in the system, synchronized with each other, taking full advantage of system resources. It is shown that real-time capability is achieved due to the multithread framework.

The result of this study has the potential to promote food safety and quality by providing advanced and automated detection techniques to the poultry and food industries.