Automobile plate number items neural based recognition methodology

In the paper, the methodology of vehicle registration number recognizing, implementing several stages, is considered. At the initial stage, the horizontal and vertical image projections color intensity analysis results are outlined in the fragments containing the vehicle registration number elements. Subsequently, the steps are carried out to identify the vehicle registration number horizontal boundaries angle of inclination using the Sobel operator and the Hough transformations, and if the vehicle image has a relatively poor quality, the adaptive binarization is implemented. In the final stage, the pattern recognition of the vehicle registration number elements is performed using a Hamming neural network.

The article describes the basic steps for using a Java-based application that implements the proposed vehicle registration number recognition methodology. We give the commands of its main menu and guidelines for ordinary and advanced users.