Computer vision on bottle filling line

In the manufacture of products in complex industrial processes, defects that may arise increase the costs borne by companies, decrease the average life of products and result in a waste of resources. Quality control processes are essential in the industrial environment.

Automated defect detection technology has advantages over manual detection, as it adapts to any environment and operates over the long term with high accuracy and efficiency. To automate a quality control process, one of the most widely used technologies is computer vision systems.

A computer vision system generally refers to a system that extracts key features from digital images. The application of these systems in industry to automate industrial processes is considered automated visual inspection. This technology is used in a variety of industries, but most notably in the food industry, where accurate filling, lid closure, overfill and underfill inspection is directly related to product safety and quality for the consumer.

In this project, the aim was to integrate a computer vision system to detect whether all bottles have caps. Bottles without caps or with damaged caps have a direct impact on the company's costs. BAÜP's requirement was to create a reliable means of automatically detecting this defect on the filling line. To this end, a vision system was configured that could evaluate simple rules based on the orientation and presence of objects.

The first step in such a system is proper image acquisition, which has to do with capturing quality images through a camera. Resolution, illumination, background colour and conveyor belt speed are factors that directly affect image quality. Once the image is captured, different analysis algorithms classify the objects (bottles) as OK or NOT OK.
Real-time communication between the vision system and the line control system ensures that non-conforming product is quickly discarded.
In this particular application, a user interface with statistical information of the analyses has also been developed.

Computer vision on bottle filling line

Proyectos