A Fast Traffic Sign Recognition Algorithm for Gray Value Images

A fast shape recognition method is presented for detecting traffic signs in monochrome images. Based on Marr's primal sketch we used a hierarchical grouping method to link low-level features of pixels like gradient orientations to high-level features like triangles and ellipses. The method is fast, because it is based on local pixel operations instead of analyzing the whole shape. Examining the local area around a pixel for parts of geometrical shapes, results are symbolic representation which are linked back to the pixel, i.e. the pixel is part of the geometrical shape. A unique description of the geometrical shape is calculated from pixels lying on the shape.

[1]  Lutz Priese,et al.  New results on traffic sign recognition , 1994, Proceedings of the Intelligent Vehicles '94 Symposium.

[2]  Bärbel Mertsching,et al.  Verkehrsszenenanalyse in hierarchisch codierten Bildern , 1993, DAGM-Symposium.

[3]  Philipp W. Besslich,et al.  Konturorientierte Verfahren in der digitalen Bildverarbeitung , 1989 .

[4]  W. Ritter,et al.  A real-time traffic sign recognition system , 1994, Proceedings of the Intelligent Vehicles '94 Symposium.

[5]  R. Janssen,et al.  An adaptive system for traffic sign recognition , 1994, Proceedings of the Intelligent Vehicles '94 Symposium.

[6]  Marco Campani,et al.  A robust method for road sign detection and recognition , 1994, ECCV.

[7]  Dana H. Ballard,et al.  Generalizing the Hough transform to detect arbitrary shapes , 1981, Pattern Recognit..

[8]  Peter Seitz,et al.  The Robust Recognition of Traffic Signs from a Moving Car , 1991, DAGM-Symposium.