Device for Damage Detection
This project represented the development of a device for damage detection of a vehicle. Such solutions already exist but are expensive, scanning takes long and can take up much space for their construction and are also complicated to use. The main focus of this project was to make an easy to use, price efficient device, yet fully functional to detect damage on the car. An important characteristic of the device was also, that it should enable quick scanning of the vehicle, without any preparations of the car. As the device should take as less space as it can, it was based on a gate-like construction type. The main target groups of damage for detection were dents, scratches and glass damage. At the time, the main focus was exploiting and acquiring data of surfaces containing dents and also scratches on painted surfaces.
The result of the project was a prototype gate-type device that exposed car damage using deflectometry. A series of price efficient cameras were positioned around the gate in such a way, that they could capture the shape of most car models. To expose dents on the surface, illuminated stripped patterns were positioned all around the gate, with the mentioned cameras recording their reflection. The pattern deformed upon crossing a dent and therefore indicating possible damage.
An article on dent detection using machine learning with deflectometry was also published in ERK 2014.