In this paper, a real time and robust algorithm for license plate recognition at the time of red light violation is presented. The plate location algorithm is based on morphological and image projection technology. At the character segmentation stage, along with the prior knowledge the region growing algorithm is adopted. Then, at the recognition stage, a reverse logical artificial neural network (RLANN) along with features mapping for some especial characters is employed. The proposed algorithm is tested and the results show its outperformance over the state-of-art methods for recognizing license plate.