The identification of road-lines can provide valuable information for automated vehicle monitoring systems. In this paper a framework is proposed for finding out the road-lines from a given traffic image using two operational levels. In the first operational level a spatial domain feature-based image segmentation and cleaning are done to remove unnecessary information of the image. In the second operational level the refined image from the first level is processed to detect the road lines accurately using a novel technique of iterative bi-directional fuzzy regression model. The validity of fuzzy regression model is further analyzed, once road-lines are obtained.