Traffic sign detection and recognition by improving Region of Interest (ROI) division to support driver assistance system

Abstract

Traffic signs detection and recognition is a technology that allows automatic identification of traffic signs. In the implementation, it mostly uses a camera to capture video or real-time conditions. The Traffic detection and recognition system can give benefits to improve safety for drivers or passengers. Therefore, traffic signs detection and recognition system are one of the important parts in Driver Assistance Systems (DAS), besides that it is also used in video surveillance to improve traffic safety in the field of Intelligent Transportation System (ITS). In this paper, we propose a method for detecting and recognizing traffic signs by combining color and shape detection and divide the interest object to several regions for object recognition. The first step is image transformation, which aims to determine interest objects in the camera view. Such a process was using hue, saturation, value (HSV) color space, grey level, and black and white (BW) level. The next step is the filter and the segmentation process. The result of the filter and segmentation process then detects area objects by using region determination. The last step is the region of interest (ROI), which is used to recognizing sign traffic by the sum of absolute differences (SAD), and we try to improve area selection of ROI around nine parts. The results of the experiment proved that the proposed method was relatively good with the success rate for detecting objects about 97,38 %, and the true positive of recognition result about 81,67 %.