MAP-Based Single-Frame Super-Resolution Reconstruction for Character Image

LI Zhan; CHEN Qing-liang; PENG Qing-yu; ZHANG Qing-feng; LI Wei-xiang

doi:10.3969/j.issn.0372-2112.2015.01.030

您当前的位置：

首页 >

文章列表页 >

MAP-Based Single-Frame Super-Resolution Reconstruction for Character Image

更新时间：2025-07-16

- MAP-Based Single-Frame Super-Resolution Reconstruction for Character Image
- Acta Electronica Sinica Vol. 43, Issue 1, Pages: 191-197(2015)
- 作者机构：
  
  1. 暨南大学计算机科学系,广东,广州,510632
  2. 暨南大学天体测量、动力学和空间科学研究中法联合实验室,广东,广州,510632
  3. 暨南大学计算机科学系,广东,广州,510632
  4. 暨南大学天体测量、动力学和空间科学研究中法联合实验室,广东,广州,510632
- 作者简介：
- 基金信息：
- DOI：10.3969/j.issn.0372-2112.2015.01.030
  CLC： TP391.41
- Published：2015
- 稿件说明：
移动端阅览
LI Zhan, CHEN Qing-liang, PENG Qing-yu, et al. MAP-Based Single-Frame Super-Resolution Reconstruction for Character Image[J]. Acta Electronica Sinica, 2015, 43(1): 191-197.
DOI：

LI Zhan, CHEN Qing-liang, PENG Qing-yu, et al. MAP-Based Single-Frame Super-Resolution Reconstruction for Character Image[J]. Acta Electronica Sinica, 2015, 43(1): 191-197. DOI： 10.3969/j.issn.0372-2112.2015.01.030.

摘要

字符是很多图像处理应用关注的区域

增强字符图像分辨率往往能提高字符识别率.针对字符图像纵横和对角线方向纹理特征

提出了一种提取不同方向纹理及图像平滑度度量的新方法

并在最大后验概率框架下

实现了单帧图像超分辨率重建算法.算法利用自适应可调模板作为卷积核

将纹理信息引入图像先验模型

最终将超分辨率重建转化为一个目标函数最优化问题.实验表明

该算法能有效提高字符图像的识别率

且对噪声具有较好的鲁棒性.

Abstract

Characters are concerned in many image processing applications.Enhance resolutions of character images frequently lead to higher recognition rate.Considering vertical

horizontal and diagonal texture features of character images

a new image smoothness measurement as well as a new method for extracting image texture in different directions is proposed

and a single-frame super-resolution (SR) reconstruction method is implemented under the maximum a posteriori (MAP) framework.Using a self-adaptive flexible template as a convolution kernel

texture information is introduced into the image prior model.Therefore

SR reconstruction is finally converted into an objective function optimization problem.Experimental results show that the proposed algorithm effectively increases the recognition rate for character images and has preferable robustness to noise.

关键词

Keywords

references

Views

下载量

CSCD

Alert me when the article has been cited

提交

Tools

Publicity Resources

Reinforcement Learning Based Tuning-free Plug-and-Play Image Reconstruction Method for Single Photon Imaging

A Kernel Method for PET Image Reconstruction with Graph Laplacian Regularization

Regularized One-Step Dynamic Image Reconstruction in Magnetic Induction Tomography

An Efficient Algorithm for Compressed Sensing Image Reconstruction

Related Author

TIAN Ye

CHEN Shuang

FU Ying

CHAI Li

SUN Kun

SHENG Yu-xia

JIN Jing-jing

Related Institution

MIIT Key Laboratory of Complex-Field Intelligent Exploration, Beijing Institute of Technology

School of Information and Electronicsy, Beijing Institute of Technology

School of Computer Science and Technology, Beijing Institute of Technology

College of Control Science and Engineering， Zhejiang University

School of Information Science and Engineering， Wuhan University of Science and Technology

⁰