Title :
Traffic sign segmentation in natural scenes based on color and shape features
Author :
Qiong Wang ; Xinxin Liu
Author_Institution :
Sch. of Comput. Sci. & Eng., Nanjing Univ. of Sci. & Technol., Nanjing, China
Abstract :
Traffic sign detection and recognition is one of the important fields in the intelligent transportation system, and is expected to provide information on traffic signs and guide vehicles during driving. Traffic sign segmentation is the first stage in traffic sign recognition system, and segmentation results influence the recognition results. This paper presents an efficient method for traffic sign segmentation in natural scenes. Firstly, the improved RGB color space is presented to obtain the initial segmentation and get the ROI in the image. Then the contour features are extracted in the binary image for moment invariants calculation. Finally, traffic signs are segmented according to the color and shape features. Experiments with a large dataset and comparison with other approaches show the robustness and accuracy of the method.
Keywords :
feature extraction; image colour analysis; image segmentation; intelligent transportation systems; natural scenes; object detection; shape recognition; traffic information systems; ROI; binary image; color features; contour feature extraction; improved RGB color space; intelligent transportation system; natural scenes; shape features; traffic sign detection; traffic sign information; traffic sign recognition system; traffic sign segmentation; vehicle guidance; Equations; Feature extraction; Image color analysis; Image segmentation; Mathematical model; Roads; Shape; improved RGB color space; moment invariants based on boundary; shape feature; traffic sign segmentation;
Conference_Titel :
Advanced Research and Technology in Industry Applications (WARTIA), 2014 IEEE Workshop on
Conference_Location :
Ottawa, ON
DOI :
10.1109/WARTIA.2014.6976273
