TRAFFIC SIGN RECOGNITION USING COLOR FILTERING AND PARTICLE SWARM OPTIMIZATION

Abstract

With the development of autonomous Driver Support Systems, automatic detection and classification of traffic signs are becoming increasingly important. This paper presents a method of identifying traffic signs by applying the algorithm to optimize particle group more reasonable than some similar researchs, while incorporating some pre-processing steps to enhance effective identification. The input data are images that captured via a camera mounted on the car. The lapse between two consecutive receiving image is ∆t second(s). Color filter and image segmentation is used to detect the location of traffic signs. Filted color objects in input image was checked and classified (prohibition signs, warning signs, not signs) by matching binary shape template. Finally, particle swarm optimization is used to identify traffic sign. This solution was tested for two types of traffic signs (prohibition signs, warning signs) and obtained results are highly accurate (average recognition rate of 93.5%).