Abstarct: Intelligent transport systems are of great importance for the various security, economic, and social reasons of today's societies. Recognition of the license plate forms the core of intelligent traffic control systems. It is very important to monitor the movement of vehicles when leaving the country for security and economic reasons. Recognition of Iran's Transit License plate is facing a serious challenge due to its four different colors.In this thesis, a method for recognition Iran's Transit Plate is presented. In the presented method after performing the preprocessing, we apply adaptive thresholding to make different binary image.Then we apply connected component labeling and property function to find characters of license plate. The extracted attributes are given to a trained neural network with 30 valid classes and 1 invalid class. Due to the number of valid components, the exact location of the plate is indicated in the image.The proposed algorithm has been tested on 54 vehicle images of Iranian transit plates from different angles, different distances, different optical and climatic conditions, with an average accuracy of 98% in license plate localization and 88% in license plate recognition.