Andrew Wilson, Editor, [email protected]
Image degradation due to atmospheric scattering caused by haze, fog, or rain is a problem in imaging applications including airborne surveillance and airport-runway security. The effects of scattering, however, can be reduced, provided the scene geometry and a physical model of the scattering effects can be determined. A spinoff from the School of Electrical and Electronics at the University of Manchester (Manchester, UK; www.man.ac.uk), Dmist Technologies (Manchester, UK; www.dmist.com), has developed a system, Clear-Vue, based on off-the-shelf components, that uses such models to perform atmospheric correction of color video images in real time.
“Light reflected from objects in a scene is attenuated, and diffuse light scattered into the camera (known as “airlight”) contributes to image degradation,” says Nigel Allinson, a director of Dmist Technologies. “Both these effects increase in magnitude exponentially, with distant areas being brighter but with lower contrast and less saturated color.” In a forward-looking camera, the large variation from foreground to background results in a loss of contrast that is nonuniform across the image.
To compensate for this effect, Dmist has developed a system that uses known range data to compensate the red, green, and blue (RGB) pixel data in each image. Because the brightness of each of the R, G, and B channels varies differently with range, each channel must be compensated for individually. Using these data coupled to known range data, it is possible to calculate the corrected R, G, and B values for each pixel and so produce fully corrected video in natural color. To calculate this range map, the camera’s height above ground, look down angle, and field of view are used.
The Clear-Vue unit, housed in a 1U enclosure, contains custom video-rate electronics and a single-board PC. It can cope with various video standards and at full broadcast frame rates and is retrofitable to almost any existing CCTV system. Once set up, there is no need for operator intervention as all video enhancement is fully automatic.
The unit also can be supplied with more powerful video-processing algorithms that do not require any knowledge of range data, as well as a new facility that enhances scenes taken under poor lighting conditions. Units are currently on trial with various international airports, defense organizations, and public-television broadcasters.
