Drives & Controls April 2024

38 n MACHINE VISION April 2024 In the world of pharmaceuticals, packaging is paramount. Blister packs are a popular choice due to their versatility and cost-e ectiveness. The individual pockets protect each OSD (oral solid dosage) item, ensuring precise dosing and long shelf lives, and revealing any tampering. However, their e ectiveness depends on the accuracy of the packaging process. Inspection is therefore a crucial task. This quality control should detect any defects in OSD items, such as chipped or broken tablets, as well as the presence of any foreign objects, and ensure that the right medications are being packed. In particular, there should not be any mix-ups on lines that process multiple medicines. It is also vital to spot any empty pockets or OSDs that have been positioned incorrectly in the blister packaging. Any o -spec blister packs need to be discarded. Traditional quality control methods such as manual inspections often fail to guarantee the required level of precision and reliability. Furthermore, they can result in long, ine ective cycle times. Vision systems can play a key role in e ective and responsive inspection of blister packs. Such systems typically combine cameras, sensors, lighting and advanced software to capture and analyse images of the blister packs as they move along production lines. The process begins with high-resolution cameras and lights positioned to generate and capture detailed images. The number of these, and their placement, depend on the type of inspection required. Lighting techniques such as di use, direct and backlighting can be used to eliminate shadows, re ections or other image distortions. Once the images have been captured, they are processed by powerful algorithms designed to detect and analyse speci†c features, irregularities, defects and other attributes of the blister packs – and the tablets they contain. The software compares the images against prede†ned criteria to determine whether each pack meets the required quality standards. Blister packs that pass the inspection are allowed to move on to downstream operations, while those that have been agged are diverted automatically and rejected from the line. This ensures that only compliant packs make their way to the next stages. Vision systems can also provide detailed data and documentation of the inspection process, including images of the inspected blister packs, timestamps and records of any defects detected. This information can support reporting and quality auditing activities to ensure regulatory compliance. They also support traceability strategies. As the capabilities of AI (arti†cial intelligence) systems evolve, vision systems are likely to become even more complex, accurate and smarter. For example, they can integrate advanced spectral functions, pattern-recognition algorithms and dataprocessing functions. They can often detect subtle anomalies that might not otherwise be identi†able. Thanks to progress in unsupervised machine learning and deep learning, creating accurate vision systems is becoming quicker. This streamlines projects and helps end-users to bene†t from quick ROIs (returns on investment). Additionally, the development of innovative 3D vision technologies will allow vision systems to provide even more detailed and accurate quality control, particularly for complex blister pack designs. n Advanced vision systems can transform blister pack inspections, making them accurate, precise and e cient. Matt Jones, account manager at Optimal Industrial Automation, looks at how they work. Machine vision systems can enhance quality control, improving speed and accuracy Vision systems check that blister packs are healthy Every blister pack and its contents must be inspected to guarantee that it meets quality standards and complies with regulatory requirements.