The license plate recognition system based on improved algorithm

Abstract

The research of this paper is based on the license plate location algorithm and Radon transform. The license plate location algorithm applies edge detection and morphological algorithm. The principle of Radon transform is adopting binarization processing for the selected global threshold and morphological processing for the license plate to segment the connected region, on the basis of the vertical projection of local areas and the character segmentation of connected areas. At the end of this paper, a character recognition system based on improved BP neural network is taken. It can overcome some difficulties that traditional algorithm confronts, such as slow in getting the optimal solution and easy to fall into the local minimum. Also, compared with existing methods, this method has a fast convergence rate and recognition speed and a high accuracy.