In this paper, a car license plate recognition (CLPR) system is presented. A CLPR system has three main subsystems such as; localization of the license plate in a given image; character segmentation of the license plate; recognition of the detected characters. Firstly, fuzzy c-means clustering is employed to determine the location of the license plate. Secondly, smearing algorithms (namely horizontal and vertical projections of the pixels of the license plate image) is used to carry out character segmentation. Finally, a novel Equipartition of Kinetic Energy (EKE)-Poisson transform is developed to recognize the characters. According to results, the proposed CLPR system is capable of supplying more than 92% of overall success rate.