Title :
A novel detection and recognition system for Chinese traffic signs
Author :
Zhang Hanxizi ; Wang Bo ; Zheng Zhihui ; Dai Yujie
Author_Institution :
Sch. of Autom., Beijing Inst. of Technol., Beijing, China
Abstract :
In this paper, a novel detection and recognition system using an on-board single camera for Chinese traffic signs is proposed. First of all, image preprocessing method using white balance is adopted to weaken the effect of illumination changes. In the detection stage, sign candidates within Regions of Interest (ROIs) are detected by the combination of color and shape information. Then the uniform rotation-invariant Local Binary Patterns (LBP) and Support Vector Machine (SVM) are applied for the recognition stage. We have conducted lots of experiments in real road scenes and the results show that our system has high accuracy and good performance in clustered backgrounds.
Keywords :
feature extraction; image colour analysis; lighting; object detection; object recognition; road traffic; support vector machines; traffic engineering computing; Chinese traffic sign; LBP; ROI; SVM; clustered backgrounds; color information; illumination changes effect; image preprocessing method; on-board single camera; regions-of-interest; road scenes; shape information; support vector machine; traffic sign detection system; traffic sign recognition system; uniform rotation-invariant local binary patterns; white balance; Image color analysis; Image segmentation; Lighting; Roads; Shape; Support vector machines; Vehicles; Detection and recognition system; LBP; SVM; Traffic signs; White balance;
Conference_Titel :
Control Conference (CCC), 2013 32nd Chinese
Conference_Location :
Xi´an
