Traffic sign recognition based on color, shape, and pictogram classification using support vector machines

Abstract

Traffic sign recognition is the second part of traffic sign detection and recognition systems. It plays a crucial role in driver assistance systems and provides drivers with crucial safety and precaution information. In this study, the recognition of the TS is performed based on its border color, shape, and pictogram information. This technique breaks down the recognition system into small parts, which makes it efficient and accurate. Moreover, this makes it easy to understand TS components. The proposed technique is composed of three independent stages. The first stage involves extracting the border colors using an adaptive image segmentation technique that is based on learning vector quantization. Then, the shape of the TS is detected using a fast and simple matching technique based on the logical exclusive OR operator. Finally, the pictogram is extracted and classified using a support vector machines classifier model. The proposed technique is applied on the German traffic sign recognition benchmark and achieves an overall recognition rate of 98.23%, with an average computational speed of 30 ms.