HIGHWAY SCENE ANALYSIS FROM A MOVING VEHICLE UNDER REDUCED VISIBILITY CONDITIONS
Margrit Betke, Huan Nguyen · 1998
This paper describes and evaluates a real-time vision system that detects and tracks vehicles on highways under reduced visibility conditions. The system uses a forward looking video camera in a moving car to estimate the distances of the car to other vehicles on the road. Results of experiments on images sequences in di#cult visibility conditions are presented. 1 Introduction Vision systems that automatically analyze highway scenes from a moving vehicle have been remarkably successful in fair weather conditions #see, for example, #2, 3, 4, 6, 7, 8, 11, 12, 13# #. Reliability of these systems, however, is also required for reduced visibility conditions that are due to rainy or snowy weather, tunnels and underpasses, and driving at night, dusk and dawn. Changes in road appearance due to weather conditions have been addressed for a stationary vision system that detects and tracks vehicles, for example, in Ref. #10#. Reference #9# describes how visibility estimates lead to reliable lane...