P. García-Díaz, S. Maldonado-Bascón, S. Lafuente-Arroyo, and R. Sánchez-Mentero (Spain)
Traffic sign, segmentation, edge detection, Hough Transform, straight-line segment, occlusion
This paper presents a new method for traffic sign classification. Traffic signs are usually classified by their geometric shape (triangular, circular, octagonal and rectangular), where each shape has specific meanings in a driving context: danger, duties, forbidden actions, information, etc. The document explains step by step the process to detect triangular signs using the Hough Transform. The algorithm can be adapted to circular and rectangular geometries. The Hough Transform is a common technique for lines extraction in images processing. A previous process is performed before calculating the transform. It mainly consists of a segmentation according the color and edge detection. This method approaches successfully problems of partial occlusions, rotations and array of traffic signs. Some of these issues are not solved in the literature. The algorithm has a low computational cost. All examples and figures are obtained from images published on www.roadanalysis.com.
Important Links:
Go Back
