In Taiwan, there are a large amount of cars in this small island, and also it comes millions of tickets because of exceeding the speed limit. Until now, the license plate is recognized manually; it costs a lot of time and manpower. License plate recognition is a powerful method that uses optical character recognition to read the characters of the plate on the car. The purpose of our research is to develop a system that can detect the speed of vehicle. If the vehicle speed exceeds the limit, then the license plate of the car will be located and the characters of the plate are also segmented. Lots of images on the streets were used to verify the proposed methods. It was found that the error of the vehicle speed calculation is less than 4km/h. License plates were locates with a correction rate of 95% and the correction rate of character segmentation is about 96%.