激光与光电子学进展, 2019, 56 (23): 231010, 网络出版: 2019-11-27
双判别器生成对抗网络图像的超分辨率重建方法 
下载： 1332次
Image Super-Resolution Reconstruction Method Using Dual Discriminator Based on Generative Adversarial Networks
图 & 表
图 3. 反卷积上采样结构的可视化棋盘伪影图像
Fig. 3. Visualized checkerboard artifact images of deconvolution upsampling structure
图 4. DDSRRN与SRGAN模型在DIV2K验证集上的R PSN与S SIM表现对比。(a) R PSN值;(b) S SIM值
Fig. 4. Comparison of S PSN and S SIM between DDSRRN and SRGAN models on DIV2K validation set. (a) S PSN;(b) S SIM
图 5. DIV2K数据集中“0846”建筑物4×重建对比图
Fig. 5. Comparison of ”0846” building reconstruction with magnification of 4 in DIV2K dataset
表 1各模型在基准数据集和DIV2K验证集上的平均性能比较
Table1. Average performance of each model on baseline dataset and DIV2K validation set
| ||||||||||||||||||||||||||||||||||||||||||||||
袁飘逸, 张亚萍. 双判别器生成对抗网络图像的超分辨率重建方法[J]. 激光与光电子学进展, 2019, 56(23): 231010. Piaoyi Yuan, Yaping Zhang. Image Super-Resolution Reconstruction Method Using Dual Discriminator Based on Generative Adversarial Networks[J]. Laser & Optoelectronics Progress, 2019, 56(23): 231010.
PDF全文
