Early on, the importance of date and lot codes was recognised to yield cost effective product recalls when necessary, however, simply applying date and lot codes was not enough, given character quality concerns associated with then existing imprinting processes. Machine vision became the ideal technology to verify codes and validate legibility of the codes. While some of the early primitive machine vision algorithms provided some guarantee of code correctness and quality, machine vision systems today have become more sophisticated with the ability to handle more compute-intensive algorithms, allowing date and lot code verification to become even more reliable.
The pharmaceutical industry was one of the first to embrace colour-based machine vision. This was driven by the need to verify that the correct birth control tablets were in the specific location in the blister package. The colours that were used were challenging for machine vision systems because they were often subtle intensity variations of the same hue. Other challenging colour-based machine vision applications in this industry included inspecting for 'foreigners' in solid dosage manufacturing and detecting particles in solutions. The very early 'machine vision' versions were based on analog techniques. As machine vision technologies improved, these analog systems were quickly replaced with digital imaging-based systems, or machine vision systems.
Read the full article here
