| 改进的二阶龙格-库塔超分辨率算法 |
| |
| 引用本文： | 陈剑涛,黄德天,陈健,朱显丞.改进的二阶龙格-库塔超分辨率算法[J].华侨大学学报(自然科学版),2022,43(1):127-134. |
| |
| 作者姓名： | 陈剑涛 黄德天 陈健 朱显丞 |
| |
| 作者单位： | 华侨大学 工学院, 福建 泉州 362021 |
| |
| 基金项目： | 国家自然科学基金青年科学基金资助项目（61901183）；;福建省自然科学基金面上资助项目（2019-J01083）； |
| |
| 摘 要： | 提出一种改进的二阶龙格-库塔超分辨率算法.首先,提出一种浅层共享编码器,以实现低分辨率图像的浅层特征提取.其次,提出一种深层特征学习单元,并与基于龙格-库塔方法的残差模块相融合,进而构建出一种基于深层特征的残差模块,以提升深层特征提取能力.实验结果表明:与主流超分辨率算法相比,文中算法在主观视觉效果和客观评价指标方面都具有更好的效果.
|
| 关 键 词： | 超分辨率 卷积神经网络 共享编码器 深度特征 |
Improved Second-Order Runge-Kutta Super-Resolution Algorithm |
| |
| CHEN Jiantao,HUANG Detian,CHEN Jian,ZHU Xiancheng.Improved Second-Order Runge-Kutta Super-Resolution Algorithm[J].Journal of Huaqiao University(Natural Science),2022,43(1):127-134. |
| |
| Authors: | CHEN Jiantao HUANG Detian CHEN Jian ZHU Xiancheng |
| |
| Affiliation: | College of Engineering, Huaqiao university, Quanzhou 362021, China |
| |
| Abstract: | An improved second-order Runge-Kutta super-resolution algorithm is proposed. Firstly, a shallow shared encoder is proposed to extract the shallow feature of low-resolution images. Secondly, a deep feature learning unit is proposed and further integrated with the residual module based on the Runge-Kutta method to construct a deep-feature-based residual module to improve the ability of deep feature extraction. Experimental results show that compared with the mainstream super-resolution algorithm, the algorithm proposed in this paper has better effect in subjective visual effect and objective evaluation index. |
| |
| Keywords: | super-resolution convolutional neural network shared encoder deep feature |
| 本文献已被 万方数据 等数据库收录！ |
| 点击此处可从《华侨大学学报(自然科学版)》浏览原始摘要信息 |
|
点击此处可从《华侨大学学报(自然科学版)》下载全文 |
|
