Seeing CLEARly through turbulence
Atmospheric distortions, due to fog or haze, or turbulent mixing due to temperature variations can severely influence the visual quality of content acquired. For example, temperature variations can change the interference pattern due to light refraction causing unclear, unsharp and ‘wobbly’ images of the objects in the scene. These effects make the acquired imagery difficult to interpret or classify.
BVI researchers Pui Anantrasirichai, David Bull and Alin Achim have devised a novel and powerful method (CLEAR) for mitigating the effects of such atmospheric distortions, particularly the difficult case of airborne turbulence. In order to provide accurate details from objects behind the distorting layer, the least distorted regions are selected, and spatially aligned. The distortion problem is then solved using region-based video fusion. The combined result has been internationally recognised as delivering state-of-the-art performance and commercial exploitation is underway with a major US industry partner.