Wheel type detection with VisionTools V60
Vision system for optical type recognition of wheels
At the end of a wheel production line, the entire range of parts is fed to a packaging system via a conveyor belt. In order to separate the wheels by type into the packaging cells, the chaotic incoming wheel variants are recognised by a camera system and subsequently sorted.
The system operates with three cameras in order to capture the different type features. The image recording is triggered by a light barrier. The wheels are inspected in a continuous process without stopping in front of the cameras.
In order to deliver a result as fast as possible, the used project software V60 gradually narrows down the possible wheel variants. If the wheel width (J dimension) and wheel diameter (H dimension) are detected, only a handful of possibilities remain, which are to be distinguished by the image processing through unique designs. The VisionTools V60 image analysis software works according to the principle of the closest match of geometric features to the inspection part present. The type with the least deviations is considered to be a match, as long as the deviations are within the set error thresholds.
In the case of an emergency wheel, the presence of the required label can also be checked. If no warning label would be necessary, the system operator simply deselects the function via checkbox and password entry.
The evaluation result with the identified type number is transferred to the control system, which then controls the switches and flaps on the subsequent transport route in order to convey the wheels to the correct packaging cell. The camera system only has one second left until the wheel arrives at the first sorting point.
The camera system can manage any number of variants. By the used method of evaluation, the recognition is very fast even though there is a large number of variants.
When new variants are added to the system, the image analysis software checks the uniqueness of the user-selected differentiating features against the existing image collections of existing variants to avoid subsequent misdetections.