论文阅读 [TPAMI-2022] Densely Residual Laplacian Super-Resolution

论文阅读 [TPAMI-2022] Densely Residual Laplacian Super-Resolution

论文搜索(studyai.com)

搜索论文: Densely Residual Laplacian Super-Resolution

搜索论文: http://www.studyai.com/search/whole-site/?q=Densely+Residual+Laplacian+Super-Resolution

关键字(Keywords)

Laplace equations; Feature extraction; Computer architecture; Convolutional neural networks; Image restoration; Super-resolution; laplacian attention; multi-scale attention; densely connected residual blocks; deep convolutional neural network

机器视觉

超分辨重建; 图像/视频修复

摘要(Abstract)

Super-Resolution convolutional neural networks have recently demonstrated high-quality restoration for single images.

最近，超分辨率卷积神经网络已经证明了对单个图像的高质量恢复。.

However, existing algorithms often require very deep architectures and long training times.

然而，现有的算法通常需要很深的体系结构和很长的训练时间。.

Furthermore, current convolutional neural networks for super-resolution are unable to exploit features at multiple scales and weigh them equally or at only static scale only, limiting their learning capability.

此外，目前用于超分辨率的卷积神经网络无法在多个尺度上利用特征并对其进行相等或仅在静态尺度上进行加权，从而限制了其学习能力。.

In this exposition, we present a compact and accurate super-resolution algorithm, namely, densely residual laplacian network (DRLN).

在本文中，我们提出了一种紧凑而精确的超分辨率算法，即稠密剩余拉普拉斯网络（DRLN）。.

The proposed network employs cascading residual on the residual structure to allow the flow of low-frequency information to focus on learning high and mid-level features.

所提出的网络在残差结构上采用级联残差，以允许低频信息流集中于学习高层和中层特征。.

In addition, deep supervision is achieved via the densely concatenated residual blocks settings, which also helps in learning from high-level complex features.

此外，通过密集串联的剩余块设置实现深度监控，这也有助于从高级复杂特征中学习。.

Moreover, we propose Laplacian attention to model the crucial features to learn the inter and intra-level dependencies between the feature maps.

此外，我们还提出了拉普拉斯注意对关键特征进行建模，以了解特征映射之间的层间和层内依赖关系。.

Furthermore, comprehensive quantitative and qualitative evaluations on low-resolution, noisy low-resolution, and real historical image benchmark datasets illustrate that our DRLN algorithm performs favorably against the state-of-the-art methods visually and accurately…

此外，对低分辨率、噪声低分辨率和真实历史图像基准数据集的全面定量和定性评估表明，我们的DRLN算法在视觉和准确度上优于最先进的方法。。.

作者(Authors)

[‘Saeed Anwar’, ‘Nick Barnes’]