A ROBUST LICENSE PLATE RECOGNITION METHODOLOGY BY APPLYING HYBRID ARTIFICIAL TECHNIQUES

摘要

Intelligent Transportation System (ITS) has been more and more important around the world. License Plate Recognition (LPR) is one of the important technologies for ITS. Meanwhile, LPR provides Real Time and Value-added services with the technology of Cloud computing and database. License plate was used to be recognized by human; however, it is tedious work and easy to make mistakes. The purpose of this research is the development of a robust license plate recognition methodology by applying hybrid artificial techniques including coordinating the angle correction algorithm and the combination of connected component theory. Meanwhile, the Optical Instrument is used as the source of images. The results reveal that about 98% plate images can be recognized by the proposed methodology after angle adjusted, but it takes more time to process the recognition without auxiliary coordinates. The successful plate recognition reached 98% when the image in a horizontal position. However, the successful rate may be greatly reduced when the angle is oblique and without any adjusted. With auxiliary coordinates, the recognition percentage is 95% when angle slope ranges are between 350° and 370°.