Road speed sign recognition using edge-voting principle and learning vector quantization network

This paper presents an automatic speed sign detection and recognition for providing the visual driving-assistance of speed limits awareness. To reduce the influence of digital noise caused by lighting condition and pollution, a segmentation based on pan-red color information is applied to extract the shape of speed sign. Based on the edge-phase information of a circle shape, a novel edge-voting principle is proposed for fast detecting the speed sign candidate from road scenes. The recognition of the content of speed sign is achieved through a modified learning vector quantization (LVQ) network which also verifies each candidate to eliminate nontarget blobs. Results show a high success rate and a low amount of false positives in both detection and recognition strategy under a wide variety of visual conditions.