Deep Networks for Degraded Document Image Binarization through Pyramid Reconstruction
Gaofeng Meng; Kun Yuan; Ying Wu; Shiming Xiang; Chunhong Pan
2017
会议名称International Conference on Document Analysis and Recognition (ICDAR)
页码727-732
会议日期November 13-15, 2017
会议地点Kyoto, Japan
摘要
Binarization of document images is an important
processing step for document images analysis and recognition.
However, this problem is quite challenging in some cases because
of the quality degradation of document images, such as
varying illumination, complicated backgrounds, image noises
due to ink spots, water stains or document creases. In this
paper, we propose a framework based on deep convolutional
neural-network (DCNN) for adaptive binarization of degraded
document images. The basic idea of our method is to decompose
a degraded document image into a spatial pyramid structure
by using DCNN, with each layer at different scale. Then the
foreground image is sequentially reconstructed from these layers
in a coarse-to-fine manner by using deconvolutional network.
Such kind of decomposition is quite beneficial, since multiresolution
supervision information can be directly introduced into
network learning.We also define several loss functions about label
consistency and foregrounds smoothing to further regularize the
training of the network. Experimental results demonstrate the
effectiveness of the proposed method.
语种英语
文献类型会议论文
条目标识符http://ir.ia.ac.cn/handle/173211/15339
专题模式识别国家重点实验室_先进数据分析与学习
推荐引用方式
GB/T 7714
Gaofeng Meng,Kun Yuan,Ying Wu,et al. Deep Networks for Degraded Document Image Binarization through Pyramid Reconstruction[C],2017:727-732.
条目包含的文件 下载所有文件
文件名称/大小 文献类型 版本类型 开放类型 使用许可
ICDAR2017.pdf(16012KB)会议论文 开放获取CC BY-NC-SA浏览 下载
个性服务
推荐该条目
保存到收藏夹
查看访问统计
导出为Endnote文件
谷歌学术
谷歌学术中相似的文章
[Gaofeng Meng]的文章
[Kun Yuan]的文章
[Ying Wu]的文章
百度学术
百度学术中相似的文章
[Gaofeng Meng]的文章
[Kun Yuan]的文章
[Ying Wu]的文章
必应学术
必应学术中相似的文章
[Gaofeng Meng]的文章
[Kun Yuan]的文章
[Ying Wu]的文章
相关权益政策
暂无数据
收藏/分享
文件名: ICDAR2017.pdf
格式: Adobe PDF
所有评论 (0)
暂无评论
 

除非特别说明,本系统中所有内容都受版权保护,并保留所有权利。