Acta Optica Sinica, Volume. 40, Issue 20, 2010001(2020)
Multi-Band Image Synchronous Super-Resolution and Fusion Method Based on Improved WGAN-GP
ing at the problem that the fused results of low resolution source images are not good for the subsequent target extraction, a multi-band image synchronous super-resolution and fusion method based on Wasserstein generative adversarial network with gradient penalty (WGAN-GP) is proposed. Firstly, the multi-band low-resolution source images are enlarged to the target size respectively based on the bicubic interpolation method. Secondly, the enlarged results are input to a feature extraction (encoding) network to extract features respectively and combine them in a high-level feature space. Then, the initial fused images are reconstructed by decoding network. Finally, a high-resolution fused image is obtained through a dynamic game between the generator and the discriminator. The experimental results show that the proposed method can not only achieve multi-band images super-resolution and fusion simultaneously, but also the information amount, clarity, and visual quality of the fused images are significantly higher than other representative methods.
Get Citation
Copy Citation Text
Songwang Tian, Suzhen Lin, Haiwei Lei, Dawei Li, Lifang Wang. Multi-Band Image Synchronous Super-Resolution and Fusion Method Based on Improved WGAN-GP[J]. Acta Optica Sinica, 2020, 40(20): 2010001
Category: Image Processing
Received: Apr. 9, 2020
Accepted: Jul. 3, 2020
Published Online: Oct. 17, 2020
The Author Email: Lin Suzhen (lsz@nuc.edu.cn), Lei Haiwei (lsz@nuc.edu.cn)