To test LED displays we still rely on a human interference to verify broken LEDs or other abnormalities in the devices with no visual proof of the outcome. Customers tend to ask to eliminate this factor as much as possible. With image processing the tests can be run without the presence of a human operator and these results can be stored as proof towards the customer.
As we run a project business we develop devices in all types and sizes. There can be a deviation in size, color, brightness, pitch,… The brightness factor can definitely be a challenge, because a lot of camera’s tend to saturate in case there is a high brightness.
Your task will be to study different types of image processing (template matching, feature detectors, cascade classifier,…). Once the most relevant one has been chosen, you can determine which hardware/software tooling makes it possible to implement this kind of processing. If a complete study is made there will be an implementation phase where a proof of concept can be developed.
Nature of the work
Specialty: Image processing
Expected outcome: proof of concept, research study