A Parallel System for Real-Time Traffic Sign Recognition

We present a system for the real-time recognition of traffic signs from a moving car on European highways. The traffic sign recognition system (TSR) was developed within the European PROMETHEUS project in cooperation with Daimler-Benz and is installed in an autonomous car. Our TSR is also intended to serve as a driver assistance tool. The TSR is based on a fast color image analysis. This analysis involves different methods, such as an inherently parallel color image segmentation, a data-driven decision graph with fuzzy techniques, and classical pattern recognition. Due to the good quality of the color segmentation and the fault-tolerant evaluation the system is highly robust against the difficult conditions in natural outdoor scenes. To meet real-time constraints the TSR has been implemented on a high speed parallel image processing system with MPC 601 processors. A prototype of the TSR runs in a usual car and reaches a recognition rate of 98 %