Design and Implementation of Deep Learning Based License Plate Recognition System

With the rapid development of intelligent transportation system, license plate recognition technology, as one of the key technologies, its accuracy and robustness are of great significance to traffic management. This paper proposes a license plate recognition system based on deep learning, aiming at the performance limitations of the existing license plate recognition technology under the influence of environmental factors such as lighting, angle, and site. The system first preprocesses the license plate image, including graying, mean filtering, edge detection and binarization, to improve image quality and reduce noise. Then, the license plate is located by combining geometric and color features, the character boundary area is determined by vertical projection method, and the characters are cut. Finally, by building a convolutional neural network (CNN) model, train and recognize license plate characters. The test and simulation in MATLAB environment show that the system identification accuracy reaches 98.6%, which verifies the effectiveness of the proposed method.