首页|基于多层感知机的ATM英文凭单识别

基于多层感知机的ATM英文凭单识别

扫码查看
文中提出一种面向英文ATM凭单的新型图像嵌入模板的处理思路,并结合机器学习进行OCR识别.首先将待识别的英文ATM凭单切割为单行图片,将此单行图片嵌入到自定义矩形模板,并将获得的单个字符图像制作成ATM凭单数据集.为了更好地识别上述处理的ATM凭单,文中采用多层感知机进行训练.经过大量实验以及对比研究,英文ATM凭单的单个字符识别准确率达到98.87%,单行识别准确率也达到了 98%.
English ATM voucher identification based on multilayer perceptron
In this paper,a new method of image embedding template for English ATM voucher is proposed,and the OCR recognition based on machine learning is carried out.Firstly,the English ATM voucher to be recognized is cut into a single line image,which is embedded into a customized rectangle template,and the single character image is made into an ATM voucher data set.To better identify the ATM vouchers pro-cessed above,this article uses multilayer perceptron for training.After a lot of experiments and comparative studies,the accuracy of single character recognition of English ATM voucher reaches 98.87%,and the ac-curacy of single line recognition is 98%.

ATM vouchersOCR recognitionimage embeddingcustomized rectangular templatemulti-layer perceptron

王文尚、高凡、侯冰莹、王林琳、冯佑

展开 >

华中师范大学物理科学与技术学院,武汉 430079

ATM凭单 OCR识别 图像嵌入 自定义矩形模板 多层感知机

2024

信息技术
黑龙江省信息技术学会 中国电子信息产业发展研究院 中国信息产业部电子信息中心

信息技术

CSTPCD
影响因子:0.413
ISSN:1009-2552
年,卷(期):2024.(5)