High Speed Real Time Machine Vision: The Canned Tuna Case

摘要

We describe a PC-based real time machine vision system for the quality assurance of canned tuna at a tin rate over 1,000 tins per minute, allowing the inspection of the tin and its contents, with two main purposes: preventing defective tins from damaging the closure equipment or from reaching the market with a defective closure, and deviating from the production line tins with low quality content. These defects include bruised or darkened areas, excessive percentage of crumb, presence of strange bodies, or partially filled tins, such that they may be fixed and reintroduced in the line without wasting any oil or sauce. The system comprises a iPentium PC with frame grabber, a grey scale line scan CCD camera, a fibre optic halogen lighting system, and a compressed air system for the deviation of the rejected tins, and has been entirely programmed in C++.