Vehicle Number Plate Identification Using a Bi-Step Region Segmentation and Classification Technique

Abstract

Vehicle number plate identification (VNPI)  is an imperative task for resolving the increasing traffic  issues around the world. Although many studies were  conducted in the past, there are still many challenges to be  answered where noisy image acquisition conditions,  improper illumination or poor quality images, are a few to  name. In the light of the same, an efficient vehicle number  plate classification model is the need of the hour. Since,  image processing techniques are best suited for resolving the  problems of noisy dataset; these are used for noise  elimination, image segmentation, feature extraction, and  classification purposes in this research. So, in this article, a  two-step approach, using region based segmentation and  feature extraction to feed as input to the system for  classifying the vehicle number plates, has been designed.  The proposed bi-step VNPI model very well extracted the  segments around the characters with extraction rate of  96.69% and recognition rate of 95.34%. Experimental  results show that the proposed technique is simple and  robust. The results are comparable with the results of the  state-of-art methods available in the literature.