基于深度学习框架的银行卡识别系统的设计

Design of Bank Card Recognition System Based on Deep Learning Framework

为了减少手工输入银行卡号的错误,提高工作效率,文章基于深度学习技术和图像处理方法设计一个银行卡号自动识别系统。系统首先通过图像处理方法对银行卡图片进行预处理,定位卡号行位置,并采用主流TensorFlow深度学习框架构建深度卷积神经网络模型进行银行卡字符的提取和识别,最后利用PyQt5进行GUI界面搭建,实现银行卡号的输出可视化。文章通过在识别模型中加入Dropout技术提高模型的鲁棒性和泛化能力,实验表明,所设计的系统能够有效地识别银行卡号。

To reduce errors in manually inputting bank card number and raise working efficiency,this paper designs a bank card number automatic recognition system based on Deep Learning technology and image processing methods.The system firstly preprocesses the bank card image by using image processing methods,locates the position of the card number line,and uses the mainstream Deep Learning framework of TensorFlow to construct a deep Convolutional Neural Networks model for extracting and recognizing bank card characters.Finally,it uses PyQt5 to build a GUI interface to achieve visual output of the bank card number.This paper improves the robustness and generalization ability of the recognition model by incorporating Dropout technology.The experiment shows that the designed system can effectively recognize bank card number.