An real-time roadside sign recognition scheme for mobile probing cars with smart phones

In recent years, intelligent traffic system (ITS) and telematics system have become a popular issue for traffic safety, convenience and efficiency. The real-time and historical traffic and other roadside information is very helpful information and can be analyzed to give good telematics services and traffic controlling, such as navigation, traffic flow, weather, traffic prediction and so on. This paper proposes an efficient real-time image recognition scheme to recognize roadside signs, such as traffic signs and common commercial marks, to collect real-time roadside traffic and other information by smart phones in mobile probing cars. The proposed scheme contains two stages. The first stage is used to detect regions of interest (ROI) which contain interesting object target, and the second stage is to recognize the detected regions. The recognition results can be collected in a cloud-based database to be further analyzed and applied in traffic information service. The experiment performs the proposed sign recognition method using traffic speed limitation signs to show its performance. The average speed is 0.085 seconds for a video frame and the average accuracy can achieve about 98%. Hence, the proposed scheme can keep lower computing complexity, however it still can obtain a well accuracy.