Automatic detection and recognition of traffic road signs for intelligent autonomous unmanned vehicles for urban surveillance and rescue

In this paper, we propose a system that automatically detects and recognizes road signs found in the United States, in real time or close to real-time. The proposed system has application to intelligent autonomous unmanned vehicles for urban surveillance and rescue. It is a multi-layered hierarchical scheme composed of 3 parts: road sign color segmentation, shape recognition, and classification. The system is robust and is invariant to image translation, rotation and scaling. It can deal with situations where there is partial occlusion, blurring of the image, and low visibility due to either weather or a change in lighting conditions. The road sign shape detection and sign classification/recognition are both based on the Principle Component Analysis. We show that the proposed system has correct classification rate of 99.2%. Experimental results show that with the current system, using existing standard hardware/software, it takes on average 2.5 seconds to detect, to segment, and to classify/recognize road signs in a road image scene. This is considered relatively fast. This time can easily be decreased in the future with dedicated specialized hardware and optimized software, taking advantage of the latest embedded hardware technology. Currently, in this paper the focus is on red and yellow road signs found in the United States but the proposed techniques can be generalized to be used for any other colored road signs found both in the United States of America and other countries.