摘要
银行每天产生大量的纸制票据,如果人工核对这些票据需要消耗大量的人力和物力,且容易出错。文章中设计了一个银行票据图像的自动识别方案,该方案包括图像预处理、票据图像分类和票据内容识别等步骤。票据的字符识别部分采用BP神经网络完成。票据的识别内容识别包括金额、传票号或交易码等的识别。实验结果表明字符识别的准确率能够满足银行的实际需求。
Much man and money power of banks will be consumed if checking paper bills is executed by manuscript. A bank bill image recognition scheme is presented which includes image preprocessing, image classification and content of bill identification. BP neural network is adopted to recognize characters. Identification of contents of bill includes amount of money, citation number or transaction code, etc. Experimental results show that the accuracy of character recognition can meet the actual demand of the bank.
出处
《信息通信》
2016年第9期157-159,共3页
Information & Communications
基金
辽宁工业大学电信学院大学生创新训练计划
关键词
票据识别
BP神经网络
图像预处理
bill recognition
BP neural network
image pre-processing
