Detection of Selected Regulatory Traffic Signs Using Common Dashboard Camera

Nowadays, many modern vehicles contain a lot of systems for autonomous functions or driver assistance systems (traffic sign recognition). Modern vehicles contain many embedded systems and control units (also for signal and image processing), so implementation of deep learning or AI-based algorithms is possible. In older vehicles, mentioned systems missing or they are limited. In this paper is presented a design of algorithm for detection of selected regulatory traffic signs to extend the selected autonomous or intelligent functions of older vehicles. Algorithm is based on shape and color detection of key features of selected regulatory traffic signs (amplification of specified color pixels and Generalized Hough Transformation). This algorithm can be easily implemented or used in many other applications, such as for driving mobile or robotic platforms. Detection efficiency of algorithm is also evaluated in this paper.