Embedded car plate image recognition system

Abstract

This project presents a low cost embedded system license plate recognition system. The system is implemented on a 8051 microcontroller platform and recognize a picture. It consists a camera and Color LCD graphic module. Captured license plate is segmented into individual character by using histogram-based approach. Character recognition is performed using template matching, which mean the segmented individual character will be compared with the character in database. In order to speed up the image processing process, the system do not use complex algorithm such as Neural Network but use a simple algorithm such as template matching. The initial step in the recognition process is obtaining a photo of the vehicle, usually by use of a mounted camera. After this, some type of algorithm must be performed to transform an image to a string consisting of the license plate number. Usually, a general framework is: Pre-processing, Bounding, Character Segmentation, and Character Recognition. The advantages of this system are its mobility. The system is small in size and therefore can bring to anywhere. Besides, low cost also its major advantage, because its use microcontroller and low cost component.