Knowledge-based region classification for rural road area in the color scene image

Unlike well-paved highway, many rural roads do not have well-defined borders. The road surface may have patches and a wide variety of color attributes in appearance. Yet in the human view of the rural road scene, the color attribute of the road area is usually distinct from those of the off-road area. And this is the major reason that we can identify the road area in the color road scene picture without relying on the 3D cues. In this paper, we present a knowledge-based scheme for region classification that helps determining the road area in the rural road scene image. Together with the property of border continuity and smoothness, the road border can be quite reasonably determined. Successful results have been shown through experiments, which mainly rely on the knowledge-based region mapping to be described.