License Plate State Recognition based on Logo Matching and State Name String Classification

In most countries, vehicle license plates contain both alphanumeric characters and the state/province of origin. State/province recognition of the license plate provides additional information to the traffic management agency and guidance information, aiding in character segmentation and in recognition. Existing methods use the character string of the state name only. Unfortunately, in some countries, the state name is too small to be captured clearly by a low-definition video camera and may also be obstructed by the license plate frame. In this study, we analyze the characteristics of license plates and propose an issuing state recognition method based on adaptive unique logo matching (AULM) and state name string recognition (SNSR). First, the AULM approach utilizes template matching to recognize the distinguishable logos. Second, if the logo is not located, the SNSR method is applied. The new method is compared with existing practices, and the experimental results show that the proposed method achieves higher accuracy and is more suitable when using highor low-definition video images.