Traffic light recognition during the night based on fuzzy logic clustering

Abstract

Traffic light recognition in night conditions is explored throughout this paper. A system detecting suspended traffic lights in urban streets is proposed. Images are acquired by a color camera installed on the roof of a car. Fuzzy logic-based clustering provides robust color detection. Additionally, other techniques end up recognizing the traffic light state. The detection rate is quite high and the false positive proportion is really low.