A Time Efficient Model for Region of Interest Extraction in Real Time Traffic Signs Recognition System

Abstract

Computation intelligence plays a major role in developing intelligent vehicles, which contains a Traffic Sign Recognition (TSR) system for increasing vehicle safety. Traffic sign recognition systems consist of an initial phase called Traffic Sign Detection (TSD), where images and colors are segmented and fed to the recognition phase. The most challenging process in TSR systems in terms of time consumption is the detection phase. The previous studies proposed different models for traffic sign detection, however, the computation time of these models still requires improvement for enabling real time systems. Therefore, this paper focuses on the computational time and proposes a novel time efficient color segmentation model based on logistic regression. This paper uses RGB color space as the domain to extract the features of our hypothesis; this has boosted the speed of the proposed model, since no color conversion is needed. The trained segmentation classifier is tested on 1000 traffic sign images taken in different lighting conditions. The experimental results show that the proposed model segmented 974 of these images correctly and in a time less than one-fifth of the time needed by any other robust segmentation methods.