Author: S. Haseena
Abstract: The best part of accessible traffic sign detection system make use of color or shape information, but the process stays limited in regard to identify and segmenting traffic signs from a complex background . In this paper, we intend a novel graph based traffic sign detection path that contains a saliency measure stage, multi-threshold segmentation stage and a graph based ranking stage. Since the graph based ranking algorithm with specific color and saliency associates the information of color, saliency, structural and dependent relationship of nodes , it is more particular and robust than the other systems in terms of use of various illumination state ,shape rotations and scale modify from traffic sign images . Additionally, the proposed multithreshold segmentation algorithm target of all nodes with a nonzero ranking score, which can efficiently solve problems such as various lighting conditions, complex background, occlusion and so on. The results for three public traffic sign position demonstrate that out proposed approach guide the better performance than the existing state of the art technique. Additionally, the results are suitable even for images hold traffic signs that have been rotated or experienced occlusion with the images that were photographed under dissimilar weather and illumination state.
Keywords: Graph based image segmentation, Graph analysis, traffic sign detection, support vector machine
Full Issue
| View or download the full issue | PDF 75-87 |