'/%3E%3C/svg%3E){kind=small}
Some packaging defects are difficult to detect reliably with a standard 2D camera. Dented corners, bulging side walls or changes in package height can be difficult to separate from variations in print, colour, reflections or surface finish.
In these cases, 3D vision provides a more direct way of evaluating the actual shape of the package.
A recent packaging inspection application from Machine Vision Shop shows how a 3D sensor and EyeVision can be used to check package geometry directly in the production process.
Inspecting the actual shape
The 3D sensor provides a point cloud of the package. EyeVision uses this data to evaluate the areas and dimensions that are relevant for the inspection.
Typical defects can include:
- dents
- bulges
- deformed sides or corners
- incorrect package height
- dimensional deviations
The main advantage is that the inspection is based on geometry rather than appearance.
A change in print or colour may look significant in a 2D image while the package itself is still perfectly within tolerance. With 3D data, the inspection can focus on the mechanical condition of the product.
Different areas of the package can also be checked independently. Side walls, corners and top surfaces can each have their own limits, depending on which types of deformation are critical for the next production step.
Configuring the inspection in EyeVision
EyeVision uses configurable processing tools that can be combined into an inspection sequence.
These tools can be used to measure dimensions, evaluate height differences and check selected areas for deformation. The individual checks can then be combined into the required inspection result.
Depending on the application, the output can be a simple PASS or FAIL signal, individual measurement values or a combination of both.
Because the inspection is configured from standard software tools, the application remains accessible after commissioning. After a short training, the end user or machine builder can adjust tolerances, fine-tune parameters or modify inspection steps themselves.
This is also useful during troubleshooting. Individual steps and intermediate results can be checked directly in the application, making it easier to identify where a deviation originates.
Routine changes therefore do not automatically require an external vision engineer.
Choosing the 3D sensor
The sensor still has to match the mechanical and performance requirements of the application.
Working distance, field of view, measuring range, required resolution and product speed all influence the choice of 3D hardware.
EyeVision can work with different 3D sensors, which gives machine builders and integrators the freedom to select the sensor that best fits the machine rather than designing the application around one fixed hardware platform.
Stable acquisition is essential
Good 3D processing starts with stable input data.
Product position, movement and synchronization influence the quality and repeatability of the point cloud. In an inline system, product guidance, triggering and encoder synchronization therefore have to be considered as part of the complete setup.
Mechanical variation that is not controlled can otherwise appear as variation in the measurement itself.
For this reason, the sensor setup, mechanics and software should be considered together during the development of the application.
Practical support during the project
For this application, Machine Vision Shop helped with the selection of the 3D sensor and supported the customer during the realization of the project.
The customer remained responsible for building and implementing the application, with additional machine vision knowledge available when required.
This approach allows machine builders and end users to keep ownership of the system themselves. Once the application is running and the users are familiar with EyeVision, most normal adjustments can be handled internally.
Support can still be used when a new inspection requirement, a major product change or a more complex technical issue comes up.
The same approach can be applied to other products where geometry is an important quality parameter, including trays, boxes, containers and formed parts.
In these applications, 3D acquisition combined with configurable EyeVision processing provides a practical way of checking product shape directly in the production line.
