A real-time vision system for defect inspection in cast extrusion manufacturing process

The cast extrusion manufacturing process is the initial step which enables the creation of the raw materials, such as clear polypropylene film, needed for the flexible packaging printing process. The current methodology of controlling extrusion-related defect occurrences is attempted by a combination of statistical sampling and human inspection. However, due to the fact that the defects are small in size and hard to visualise in a clear thin film 2 m in width moving at a speed of 50 m/min. This results in poor product quality and high return ratio from customers. To the best of our knowledge, there is no system available that can accurately detect such defects. This research investigates possible defect detection methodologies and has subsequently proposed a system that is capable of real-time monitoring of defects on the cast extrusion manufacturing process. The proposed system utilises the refraction of a collimated light source, which is referred to as Mie light scattering. A vision analysis system is subsequently used to perform a blob analysis to detect the contrasting dark regions of the defects. Two test rigs were constructed to test the feasibility of the system. The first test rig was created to test the theoretical Mie scattering principles and the performance of the image analysis software in practice. The second test rig was created to test the practicability of integrating the Mie scattering theory on the physical cast extrusion line. The results obtained from the tests indicated a success rate of 90% in identifying gels and a 100% success rate in correctly identifying all the die lines presented in the tested samples. It is also deduced that the software has a capability to detect gel granules with a diameter greater than 480 μm and die lines with a thickness greater than 320 μm amid complete repeatability, ensuring that the proposed system fully conforms to the standard industrial requirements.