Detail-Preserving Transformer for Light Field Image Super-Resolution
About
Recently, numerous algorithms have been developed to tackle the problem of light field super-resolution (LFSR), i.e., super-resolving low-resolution light fields to gain high-resolution views. Despite delivering encouraging results, these approaches are all convolution-based, and are naturally weak in global relation modeling of sub-aperture images necessarily to characterize the inherent structure of light fields. In this paper, we put forth a novel formulation built upon Transformers, by treating LFSR as a sequence-to-sequence reconstruction task. In particular, our model regards sub-aperture images of each vertical or horizontal angular view as a sequence, and establishes long-range geometric dependencies within each sequence via a spatial-angular locally-enhanced self-attention layer, which maintains the locality of each sub-aperture image as well. Additionally, to better recover image details, we propose a detail-preserving Transformer (termed as DPT), by leveraging gradient maps of light field to guide the sequence learning. DPT consists of two branches, with each associated with a Transformer for learning from an original or gradient image sequence. The two branches are finally fused to obtain comprehensive feature representations for reconstruction. Evaluations are conducted on a number of light field datasets, including real-world scenes and synthetic data. The proposed method achieves superior performance comparing with other state-of-the-art schemes. Our code is publicly available at: https://github.com/BITszwang/DPT.
Related benchmarks
| Task | Dataset | Result | Rank | |
|---|---|---|---|---|
| Light Field Image Super-Resolution | HCIold 2x scale (test) | PSNR44.3 | 11 | |
| Light Field Image Super-Resolution | INRIA 2x scale (test) | PSNR36.41 | 11 | |
| Light Field Image Super-Resolution | Average 2x scale (test) | PSNR38.4 | 11 | |
| Light Field Image Super-Resolution | EPFL 2x scale (test) | PSNR34.49 | 11 | |
| Light Field Image Super-Resolution | STFgantry 2x scale (test) | PSNR39.43 | 11 | |
| Light Field Image Super-Resolution | EPFL 4x scale (test) | PSNR28.94 | 11 | |
| Light Field Image Super-Resolution | HCInew 4x scale (test) | PSNR31.2 | 11 | |
| Light Field Image Super-Resolution | HCIold 4x scale (test) | PSNR37.41 | 11 | |
| Light Field Image Super-Resolution | INRIA 4x scale (test) | PSNR30.96 | 11 | |
| Light Field Image Super-Resolution | STFgantry 4x scale (test) | PSNR31.15 | 11 |