PConvSRGAN: Real-world super-resolution reconstruction with pure convolutional networks

• Published in: Computer vision and image understanding Vol. 260; p. 104465
• Main Authors: Zhao, Zuopeng, Gao, Yumeng, Min, Bingbing, Miao, Xiaoran, Hu, Jianfeng, Liu, Ying, Pharksuwan, Kanyaphakphachsorn
• Format: Journal Article
• Language: English
• Published: Elsevier Inc 01.10.2025
• Subjects: Multi-scale features; Probabilistic degradation; Pure convolutional network; Super-resolution reconstruction; Pure convolutional network; Super-resolution reconstruction; Multi-scale features; Probabilistic degradation
• ISSN: 1077-3142
• DOI: 10.1016/j.cviu.2025.104465

Image super-resolution (SR) reconstruction technology faces numerous challenges in real-world applications: image degradation types are diverse, complex, and unknown; the diversity of imaging devices increases the complexity of image degradation in the super-resolution reconstruction process; SR requires substantial computational resources, especially with the latest significantly effective Transformer-based SR methods. To address these issues, we improved the ESRGAN model by implementing the following: first, a probabilistic degradation model was added to simulate the degradation process, preventing overfitting to specific degradations; second, BiFPN was introduced in the generator to fuse multi-scale features; lastly, inspired by the ConvNeXt network, the discriminator was redesigned as a pure convolutional network built entirely from standard CNN modules, which matches Transformer performance across various aspects. Experimental results demonstrate that our approach achieves the best PI and LPIPS performance compared to state-of-the-art SR methods, with PSNR,SSIM and NIQE being on par. Visualization results show that our method not only generates natural SR images but also excels in restoring structures. •Introduce a probabilistic degradation model to assist in simulating the image degradation process and provide prior knowledge of the degradation process.•Introduce BiFPN, widely applied in object detection, for multi-scale feature fusion.•Improve the discriminator to a pure convolutional network that outperforms Transformer-based models.•Utilize multi-level feature extraction and fusion techniques to maintain efficient feature extraction while fully leveraging computational resources.