Character Recognition Based on Region Pixel Concentration for License Plate Identification

, . Character recognition is a broad term and consequently its application is possible in different areas and in different ways. In license plate character recognition, the complexity of the process enables the use of different methods or even combinations of methods to solve the problem. Unpredictable nature of input data (digital images) makes it difficult to choose a universal method that would be appropriate for every input image. Often the advantages of a method in one area are also disadvantages in the other. Automatic license plate recognition is a process with the main aim of getting textual form of license plate characters from digital image of a vehicle. Digital image as an input passes through several stages of processing to obtain the string of characters as a final result. The core of this process is algorithm for optical character recognition (OCR) [1] Optical character recognition process has become very common in technical systems since the speed and performance of today's computers allow fast digital image processing. It also represents the main part of the automatic license plate algorithm implemented in this paper. Apart from the mentioned process, several other actions are necessary to perform the whole process of license plate identification. This primarily refers to image filtering, license plate area detection and image segmentation. There are many different ways to solve such problems. This paper implements some new ideas in this kind of character recognition process. The ideas consider additional threshold methods for plate detection and region pixel concentration as main sources of improvement. Our main goal is to achieve successful recognition of license plate characters with less influence of input image deformation.