摘要
文本行提取一直是手写文档图像分析与识别领域的热点研究课题。随着深度学习的发展,越来越多的方法涌现出来,通过对近几年的相关文献分析整理,本文按照全卷积神经网络、编解码器、循环神经网络、生成对抗网络的基于深度学习的手写文本行提取方法进行总结和分析,列举了每种方法的代表性实例,并对常用训练数据集进行了介绍。分析了各类方法的特点与不足,并对未来可研究方向进行展望。
Text line extraction,as a crucial step in image analysis and recognition of handwritten documents,has always been a hot research topic in this field.With the development of deep learning,more and more methods have emerged.Through the analysis and sorting of relevant documents in recent years,this article is based on deep learning based on full convolutional neural networks,codecs,recurrent neural networks,and generative adversarial networks.Summarize and analyze the handwritten text line extraction methods,list representative examples of each method,and introduce commonly used training data sets.Finally,the four types of methods listed are summarized,the characteristics and shortcomings of each method are analyzed,and the future research directions are prospected,and three suggestions are put forward.
作者
杨益暄
田益民
崔圆斌
齐千慧
韩利利
YANG Yixuan;TIAN Yimin;CUI Yuanbin;QI Qianhui;HAN Lili(Beijing Institute of Graphic Communication,Beijing 102600,China)
出处
《智能计算机与应用》
2020年第11期154-157,160,共5页
Intelligent Computer and Applications
基金
国家自然科学基金(6378001)。
关键词
手写文档图像
深度学习
手写文本行提取
handwritten document image
deep learning
handwritten text line extraction