Detection of vehicle manufacture logos using contextual information

Abstract

Besides the decorative purposes, vehicle manufacture logos can provide rich information for vehicle verification and classification in many applications such as security and information retrieval. Detection and recognition of vehicle manufacture logos are, however, very challenging because they might lack of discriminative features themselves. In this paper, we propose a method to detect vehicle manufacture logos using contextual information, i.e., the information of surrounding objects near vehicle manufacture logos such as license plates, headlights, and grilles. The experimental results demonstrate that the proposed method is more effective and robust than other methods. © Springer-Verlag 2010.