摘要
为了解决发票手工分类效率低、劳动强度大等问题,提出一种基于深度学习的发票图像分类方法,实现发票自动分类。从网络爬取相关发票图像,并对图像进行滤波、倾斜校正等预处理,然后建立发票图像数据集,用深度残差网络(ResNet网络)模型进行训练,并在测试集上进行测试,结果显示,发票图像分类准确率高,能满足分类需求。
In order to solve the problems of manual invoice classification, such as low efficiency and high labor intensity, a method for classification of invoice images based on deep learning is proposed to automate invoice classification. Grab the relevant invoice images from the Internet and preprocess the images by filtering and tilting correction etc., the invoice image data bank is established. Training is conducted by using deep residual network(ResNet) model, and test is done on the test bank. The results show that the classification of invoice images is of high accuracy which meets the classification requirements.
作者
张振亚
倪红军
ZHANG Zhen-ya;NI Hong-jun(School of Economics and Management,Nantong University,Nantong 226019,China;School of Mechanical Engineering,Nantong University,Nantong 226019,China)
出处
《南通职业大学学报》
2020年第2期79-83,共5页
Journal of Nantong Vocational University
基金
教育部产学合作协同育人项目(201901189019)
江苏省大学生创新创业训练计划项目重点项目(201910304055Z)。
作者简介
张振亚,(1998-),男,江苏丹阳人,主要研究方向为人力资源管理与智能学习;通讯作者:红军,(1965-),男,江苏南通人,博士,教授,硕士生导师,主要研究方向为新能源、固废利用与智能装备。
