Traffic sign detection, recognition and transmission system for smart vehicles
Abdelhamid Mammeri, Azzedine Boukerche, Renfei Wang, Jingwen Feng · 2013
The Traffic Sign Detection and Recognition system (TSDR) has been mainly designed to increase driver safety by the fast perception and interpretation of Traffic Signs (TSs). In almost all published TSDR systems, a two-stage strategy was adopted: a detection stage followed by a recognition stage. However, in such systems, the recognized TS is used locally and this data is not shared with neighboring vehicles. In some dangerous situations, Vehicle to Vehicle communication of TS information is required to prevent approaching drivers from a possible crash. In this paper, a novel system based on the detection, recognition and transmission of TSs is investigated. Hence, two main modules have been developed: Vision-Module (VM) and Transmission-Module (TM). In the VM, an efficient algorithm for the detection and recognition of TS is presented. A modified version of Histogram of Oriented Gradients (HOG) is used to detect and recognize TSs through a set of multi-level Support Vector Machines (SVM) classifiers. The goal of the TM, however, is to send the TS information from a vehicle detecting the TS to the following vehicles belonging to a destination area called zone of relevance. Consequently, a geocast routing protocol is simulated using NS-2. A set of tests are performed to evaluate these modules.