A real-time traffic sign recognition system based on local structure features

Abstract

We present an accurate and efficient system for traffic sign recognition in a real-world driving scene video. The proposed system uses local structure features to achieve high, illumination-invariant accuracy in detection and recognition. We exploit a property of traffic signs, namely, shared boundary shapes, to enhance the speed and accuracy of the detection step. A multi-level SVM structure is employed for stable recognition. The proposed method can process real-world road driving scene video in real time with high accuracy, over 98%, in both detection and recognition. © International Conference on Image Processing, Computer Vision, and Pattern Recognition, IPCV 2015.All right reserved.