Traffic Sign Recognition Method in Intelligent Transport System based on the Low-rank Approximation

摘要

This paper presents a new method for traffic sign recognition in Intelligent Transport System, which base on low-rank approximation and support vector machine (SVM), the method including traffic signs correction and SVM identification. First we extraction traffic sign region and internal texture, according to the characteristics of internal texture, combine with the spare and low-rank approximation, to correct the texture automatically, next to extract the feature vectors of traffic signs texture, finally identification in the database. The experimental results show: the method base on low-rank approximation can corrected the deformation traffic signs effectively and accurately, improve the recognition rate of the SVM, it has good feasibility and real-time.