Robust Automatic Traffic Signs Detection using fast polygonal approximation of digital curves

摘要

Traffic Sign Recognition is nowadays widely used in the automobile industry, it provides a good solution for driver assistance systems, it can alert the driver and control or prohibit certain actions, and this can significantly increase driving safety and comfort. This paper presents a study to design, implement and test a method of detecting traffic signs, based on computer vision. The system implemented in this work contains two main modules: the first is the preprocessing module which is based on image processing features such as color segmentation, threshold technique, Gaussian filter, canny edge detection, and contours detection. The second is the detection module based on Polygonal Approximation of digital curves technic applied on contours to identify areas of the scene may contain traffic signs.. The development of the system is performed using the programming language C/C++ with the OpenCV library. The tests are performed on a set of real images of traffic with a complex background, to show the performance of the system being developed.