University of Minnesota Driven to Discover
U of MNUniversity of Minnesota
Center for Transportation Studies

A new view of visibility on the highway

When fog or blowing snow reduces visibility on a busy highway, a single collision can escalate quickly into a deadly multiple-vehicle pile-up. Notifying motorists of low-visibility conditions in advance, so they can reduce speed and proceed safely, is key to preventing such crashes, but continuously monitoring visibility along hundreds of miles of road has seemed like an impossible task.

At the University of Minnesota’s Duluth campus, where blustery winter weather is commonplace, professor Taek Kwon of the Department of Electrical and Computer Engineering has taken on the challenge of developing an automated system to monitor visibility along highways.

The problem of developing an automated system to monitor visibility is a difficult one because visibility itself is a complex function of many atmospheric variables. The amount of light available and the size and density of airborne particles, as well as reflectivity, light absorption, and the size, shape, and color of visible objects—all these factors affect visibility. Measuring all the variables, and using them to compute a value corresponding to human perception, would be incredibly difficult and prone to errors; most visibility meters measure at most one or two atmospheric parameters and generate an index of visibility based on those.

Instead of measuring the atmospheric variables directly, Kwon’s solution to this problem was to measure the visual properties of video images, which are roughly analogous to the images received by the human eye. Using image-processing software, a computer analyzes individual video frames to derive a measure of how much information is lost, based on comparison to an image of the same scene taken under ideal conditions. This measurement is known as relative visibility or RV.

Kwon created an experimental system using a video camera and several targets mounted at fixed distances along a highway near Duluth, where he was also able to carry out initial tests of a similar approach to nighttime visibility measurement, using near-infrared cameras rather than visible-light cameras.

Atmospheric Visibility Measurements Using Video Cameras: Relative Visibility is available online.