Institut für Mess- und Regelungstechnik (MRT)

Visibility Assessment for Driver Assistance Systems

Motivation:

Future driver assistance systems (DAS) will have to focus on adverse weather conditions, since e.g. wet roads or limited visibility caused by heavy rain or fog can severely decrease traffic safety.

 

Many state-of-the-art computer vision algorithms are designed to work in well-posed visibility conditions. Though this assumption may hold for most indoor and simulation scenarios, environmental noise can badly influence their performance in outdoor settings, e.g. in surveillance applications or in driver assistance systems where a camera is mounted behind the windshield of a moving vehicle. However, proper operation in these situations is a security-relevant prerequisite to many applications, particularly on board mobile vehicles.

 

Visibility Distance Estimation:

 

In clear weather conditions, the radiance from a scene point reaches the observer unaltered. However, dealing with adverse weather conditions, atmospheric effects cannot be neglected anymore. In recent literature, the Koschmieder Model has been established as a description of the atmospheric effects of weather on the observer. Hence, brightness and contrast measures can be deduced from the Koschmieder equation above:

 

 

Rain Detection

State-of-the-art rain sensors can only estimate the rain
strength. In addition, deducing the image quality purely
from rain sensor observations is not valid.
The objective in this research field is an accurate local
detection of raindrops on windshields. Different feature
detectors are developed and combined with a
photometric raindrop model that simulates an artificial
raindrop pattern by tracing the light rays through a
potential raindrop.