Abstarct: Intelligent transport systems have been presented as an inevitable solution for traffic problems and loss of time, money and lives, the ability to identify the offending car and law enforcement surveillance on that case, is vital. Shadow removing and ambiguity elimination is One of the cases that the license plate image recognition algorithms is considered. Considering the shadow as the part of an object in shaded images, is causing problems in image processing. Thus far, techniques and different algorithms are used to detect and remove shadows. In this thesis, a pre-processing to correct non-uniformity of brightness of plate is presented. Retinex method for correcting the brightness for the first time on the Persian plates were used and acceptable results have been achieved. The output of this pre-processing application have been given to plate recognition application of Farsi Pardazeshgaran group and Golbafan group and have been improved the results. For example, in hard shadow, the probability of false detection of license plate in the Farsi Pardazeshgaran group system decreased from the 68.75% to 2.5% and in the Golbafan group system decreased from 80% to 58.75%.