Attribute Identification for License Plates with Different Style

A new, robust, and computationally efficient algorithm for the license plate attribute identification is presented. (attribute means one-row or two-row characters that a license plate has in this paper). The proposed approach provides a solution for the traditional vehicle license plate recognition system to differentiate the existing license plates with the style of "two-zero-zero-two" issued in recent years. It overcomes the defects caused by the traditional methods based on pixels projection. What it remarkably differs from the traditional methods is the application of image pruning and the calculation of maximum character pixels distance. The two steps increase the identification rate to an extent. The experimental results demonstrate the method\´s accuracy, feasibility and robustness. It would be a potentially significant contribution to the active area of license plate recognition system