| Summary: | 碩士 === 大同大學 === 資訊工程學系(所) === 98 === This paper proposes an adaptive road color model for traffic related applications such as the detection of illegal lane change and reverse driving direction. We could set up an adaptive road surface color model by utilizing the large size property to detect the road surface under variety of weather conditions, lighting, and shadow. Color and Hough transform are then used to detect different lane lines. To overcome the problem of vehicle occlusion, visual width, visual length, and corner points are used to segment each occluded vehicle. On the other hand, gaps within vehicle components are filled by finding the outlines. Color, mass center, and moving direction are adopted for vehicle tracking. Experimental results show that the detection rate of vehicles, reverse driving, and illegal lane change reach 94.8%, 94.9% and 96.4%, respectively.
|
|---|
