At Oculus Vision, we pride ourselves on being more than just a standard machine vision integrator. We are independent technical experts who tackle complex and challenging machine vision applications. We provide unbiased advice, focusing on the right solution for your specific needs. This case study demonstrates our unique approach to solving an intricate challenge: how to read strings of direct part marked text at a rate of more than 2,000 per minute.

The Challenge: Ensuring printed Codes Are Correct

In the process of producing canned food, our client manufactures and seals products in what are initially unlabeled cans. The only way to identify the contents, production date, and manufacturing line is a four-digit code printed on the underside of each can.

Ensuring this code is correct and readable by the human eye is critical for accurate labeling. However, this task is complicated by the high-speed production environment. With cans moving at over 2,000 per minute, the printed codes are often skewed, stretched, or squashed. Missing dots from the inkjet printing process also made traditional machine vision methods unreliable for this application.

Our Solution

Standard vision systems struggle with this level of speed and print variation. To overcome this, Oculus Vision developed an innovative imaging solution to acquire a stable image of the can's underside as it traveled at high velocity. Our unique method flattened the concentric, raised circles on the can's base, resulting in a clean, undistorted image of the code.

For the code reading itself, we utilised the advanced capabilities of the MVTec Halcon DeepOCR tool. This deep-learning-based software is specifically designed to read characters robustly, regardless of orientation, font type, or print quality. To handle the high throughput, this solution was only possible by leveraging a powerful GPU to process images simultaneously in batches. This parallel processing was critical to keeping pace with the rapid line speed and ensuring a reliable result for every single can.

The Outcome

By integrating our custom solution with the client's control system, any can with an unreadable or incorrect code is now automatically removed from the line.

This new process ensures that every can is correctly labeled before it leaves the factory. The result is a highly efficient and accurate system that eliminates human error, reduces waste, and guarantees product traceability at a rate of over 2,000 cans per minute.