基于神经网络和图像识别的车牌识别技术

License plate recognition technology based on neural network and image recognition

针对车辆管理信息化的进一步需求,文中基于图像识别技术对车牌识别的相关算法进行研究。使用深度学习领域的神经网络技术,对车牌识别算法进行改进。算法使用卷积神经网络结构,以提高车牌识别的准确率为目的,合理设置网络参数,通过误差的反向传播优化各层的权重。网络经过训练和测试,识别准确率达到97%以上,耗时5.2 ms,从而提高了车牌识别技术的实用性。

Aiming at the demand of vehicle management informatization in our country,this paper studies the relevant algorithms of license plate recognition based on image recognition technology.In this paper,the neural network technology in the field of deep learning is used to improve the license plate recognition algorithm.The algorithm uses convolution neural network structure to improve the accuracy of license plate recognition,set network parameters reasonably,and optimize the weight of each layer through error back propagation.After training and testing,the recognition accuracy of the network can reach more than 97%,which takes 5.2 ms,and improves the practicability of license plate recognition technology.