Abstract
Traffic sign detection and recognition plays a prominent role in advanced driver assistance systems (ADAS). Traffic signs provide information about the traffic rules, road conditions, route directions and assist the drivers for better and safe driving. This article proposes an algorithm for traffic sign detection and recognition in video frames. The detection is performed in the hue-saturation-value (HSV) color space using both color and shape features and the recognition is performed using the multilayer perceptron (MLP) trained with histogram of oriented gradients (HOG) features. The proposed method implemented on Microsoft Visual Studio-2010 using OPENCV 2.4.8.0 image processing library. Experimental results reveal that a recognition rate of 97.14% and 95.71% is achieved on real-time videos.