Automatic License Plate Recognition System for Recognizing Overlapping Characters in Malaysian License Plates

License plate is used for identification of vehicles. Automatic License Plate Recognition (ALPR) has been developed in application such as parking and toll fees, and assisting police in law enforcement. It has been captured the attention of the industries developing this technology. In Malaysia, there are difficulties for current ALPR to recognize the license plate. One of the problems encountered by existing ALPR system when recognizing Malaysian license plates is overlapping characters which the arrangement of the characters is very close to each other. This study will focus on overlapping characters with a standard font that have been set in Malaysia. Connected component analysis have been identified that can solve the problem. It can be used to segment the characters correctly for the characters recognition. This study will be performed on extracted image of license plate only without the vehicle body. It will not include the process of plate localization. This study also involves MATLAB software as the platform in developing and simulating the ALPR system. The steps of ALPR system that will be covered in this study is Pre-processing, Character segmentation and Character recognition. The algorithm has been developed for the Pre-processing part, connected component labeling and template matching. The template matching produces accuracy of 66.67% recognizing the characters. Euler number and matrix pixel values extraction technique have been added for the features extraction process. Characters that have the same features will be recognize by template matching. The system was tested with the simulated image that have been used previously and achieved 100% accuracy. More simulated image will be created and tested with variations in font. The system will also be tested with using real image captured. The test achieved 75% accuracy. Wrong recognition of license plate is due to different in lighting, skewed image and different font. More sample image need to be collected and tested into the system to discovers more flaws on the system. This would able us to further improve the system