针对用户在移动端进行银行卡卡号录入出错性高的问题，本文提出了一种基于移动端的银行卡识别方法。方法主要分为三步：银行卡图像获取及预处理，银行卡号数字字符分割，数字字符识别。其中图像预处理部分主要利用各类边缘检测算法、形态学操作进行处理，具体包括对银行卡进行投影矫正、获取卡号区域。获得数字区域后，再进行卡号提取、数字分割，最后使用卷积神经网络(CNN)训练模型，实现最终的数字识别。最后通过开发的基于Android客户端的App进行多次试验，验证了该方法能够较好地识别银行卡卡号。
In view of the problem that the user has a high error in entering the bank card number on the mo-bile terminal, this paper proposes a bank card recognition method based on the mobile terminal. The method is mainly divided into three steps: bank card image acquisition and preprocessing, bank card number digital character segmentation, digital character recognition. The image pre-processing part mainly uses various edge detection algorithms and morphological operations to process, including the projection correction of the bank card and the acquisition of the card number area. After obtaining the digital area, the card number is extracted and the number is segmented. Finally, the convolutional neural network (CNN) is used to train the model to realize the final number recognition. Finally, through the development of the app based on Android client for many times, it is verified that this method can identify the bank card number better.
Zhang, H., Zhao, K., Song, Y.-Z., et al. (2013) Text Extraction from Natural Scene Image: A Survey. Neurocomputing, 122, 310-323. https://doi.org/10.1016/j.neucom.2013.05.037
