Nowadays, automatic detection and recognition of traffic signs is one of the interest topics to researchers. Researches rely on the automatic extraction of the signs with the desirable speed and accuracy. The algorithms are developed in such a way that the system can detect and recognize traffic signs in a highly complex urban environment. A lot of research has been done on this issue in recent decades, but with this extensive research, this topic is still attractive to researchers for future research. The reason is the existence of some challenges that reduce the accuracy of existing algorithms. These are intensive changes in light condition, shadow, the existence of obstacles, scale and orientation changes, the existence of a complex field, the speed of the algorithm, and the cheapness of the system.
Generally, the process of detecting and identifying traffic signs is performed in three separate sections, consisting of image segmentation, detection and recognition of signs. In each of these processing steps, the above challenges may cause the identification process to be difficult. For example, in the segmentation step, the intense light changes of the imaging environment affect the processing results. Also, in detection step, the presence of obstacle against the signs in the accrued images cased Detection of the shape of the sign is difficult. Hence, in this paper, an overview of the existing methods for each of the three steps of detection of traffic signs and also their challenges are discussed.