A Novel Solution for the optical inspection of windshields ADAS region

Cameras mounted behind the windshield and used for ADAS (Autonomous Driving Assistance Systems) are more and more common in today's cars. The quality of the glass just before their objective is critical for these systems to work properly. Despite more than 50 different standards for the inspection of automotive glasses, it does not exist yet a reliable solution for the characterisation of the quality key parameters of these ADAS areas.

Snellium is currently developing an innovative solution for the inspection of such particular areas of windshields, combining multiple technologies measuring image sharpness, contrast and distortion. Our solution inspects the glass quality by analysing images of a pattern grabbed by the ADAS camera through the windshield.

Our Setup

The windshield inspection is a relative measurement. The effect of the windshield on the image is deduced by comparing two images: one acquired through the windshield and one without.

The mechanical setup is simple: it is composed of a single frame supporting both the windshield and the camera, allowing to place them in desired positions. A pattern is placed in front of the camera, typically at a distance of a few meters.

Once the images have been acquired, we process them using our custom software tools, characterising the windshield performances in sharpness, contrast and distortion. To know more about these processes, take a look at our presentations of Digital Image Correlation and Modulation Transfer Function algorithms.

Performance Level Classification

Once windshield properties are obtained, how to know if  the glass performances are satisfactory? From our point of view, the best way is to verify that the image obtained through the windshield is compliant with the ADAS camera algorithm.

Doing tests in real-life environment can be costly. Snellium came up with an easy and simple solution to assess the performances of ADAS software under the glass effects.

Given the results of our test campaigns, we built a simulator of the windshield: a piece of software that can degrade any image like the windshield would do. Then, the images can be sent to the ADAS software to evaluate how it behaves under the windshield effects.

To know more about this simulator, please take a look at our Degradation Model presentation.