Improving Bracket Image Restoration and Enhancement with Flow-guided Alignment and Enhanced Feature Aggregation (2404.10358v1)
Abstract: In this paper, we address the Bracket Image Restoration and Enhancement (BracketIRE) task using a novel framework, which requires restoring a high-quality high dynamic range (HDR) image from a sequence of noisy, blurred, and low dynamic range (LDR) multi-exposure RAW inputs. To overcome this challenge, we present the IREANet, which improves the multiple exposure alignment and aggregation with a Flow-guide Feature Alignment Module (FFAM) and an Enhanced Feature Aggregation Module (EFAM). Specifically, the proposed FFAM incorporates the inter-frame optical flow as guidance to facilitate the deformable alignment and spatial attention modules for better feature alignment. The EFAM further employs the proposed Enhanced Residual Block (ERB) as a foundational component, wherein a unidirectional recurrent network aggregates the aligned temporal features to better reconstruct the results. To improve model generalization and performance, we additionally employ the Bayer preserving augmentation (BayerAug) strategy to augment the multi-exposure RAW inputs. Our experimental evaluations demonstrate that the proposed IREANet shows state-of-the-art performance compared with previous methods.
- Burst image deblurring using permutation invariant convolutional neural networks. In ECCV, pages 731–747, 2018.
- Ntire 2021 challenge on burst super-resolution: Methods and results. In CVPRW, pages 613–626, 2021a.
- Deep burst super-resolution. In CVPR, pages 9209–9218, 2021b.
- Deep reparametrization of multi-frame super-resolution and denoising. In ICCV, pages 2460–2470, 2021c.
- Ntire 2022 burst super-resolution challenge. In CVPRW, pages 1041–1061, 2022.
- Self-supervised burst super-resolution. In ICCV, pages 10605–10614, 2023.
- Basicvsr++: Improving video super-resolution with enhanced propagation and alignment. In CVPR, pages 5972–5981, 2022.
- Improving dynamic hdr imaging with fusion transformer. In AAAI, pages 340–349, 2023.
- Hdr imaging with spatially varying signal-to-noise ratios. In CVPR, pages 5724–5734, 2023.
- Burst deblurring: Removing camera shake through fourier burst accumulation. In CVPR, pages 2385–2393, 2015.
- Burst image restoration and enhancement. In CVPR, pages 5759–5768, 2022.
- Burstormer: Burst image restoration and enhancement transformer. In CVPR, pages 5703–5712. IEEE, 2023.
- Hdr image reconstruction from a single exposure using deep cnns. ACM TOG, 36(6):1–15, 2017.
- Deep burst denoising. In ECCV, pages 538–554, 2018.
- Burst photography for high dynamic range and low-light imaging on mobile cameras. ACM TOG, 35(6):1–12, 2016.
- Low-light image enhancement with wavelet-based diffusion models. ACM TOG, 42(6):1–14, 2023.
- Deep high dynamic range imaging of dynamic scenes. ACM TOG, 36(4):144, 2017.
- Joint demosaicing and deghosting of time-varying exposures for single-shot hdr imaging. In ICCV, pages 12292–12301, 2023.
- Learning raw image denoising with bayer pattern unification and bayer preserving augmentation. In CVPRW, pages 0–0, 2019.
- Joint hdr denoising and fusion: A real-world mobile hdr image dataset. In CVPR, pages 13966–13975, 2023.
- Single-image hdr reconstruction by learning to reverse the camera pipeline. In CVPR, pages 1651–1660, 2020.
- Adnet: Attention-guided deformable convolutional network for high dynamic range imaging. In CVPRW, pages 463–470, 2021.
- Ghost-free high dynamic range imaging with context-aware transformer. In ECCV, pages 344–360. Springer, 2022.
- Ebsr: Feature enhanced burst super-resolution with deformable alignment. In CVPRW, pages 471–478, 2021.
- Bsrt: Improving burst super-resolution with swin transformer and flow-guided deformable alignment. In CVPRW, pages 998–1008, 2022.
- Burst denoising with kernel prediction networks. In CVPR, pages 2502–2510, 2018.
- Deep multi-scale convolutional neural network for dynamic scene deblurring. In CVPR, pages 3883–3891, 2017.
- Hdr-gan: Hdr image reconstruction from multi-exposed ldr images with large motions. IEEE TIP, 30:3885–3896, 2021.
- Ntire 2021 challenge on high dynamic range imaging: Dataset, methods and results. In CVPRW, pages 691–700, 2021.
- Towards practical and efficient high-resolution hdr deghosting with cnn. In ECCV, pages 497–513. Springer, 2020.
- Labeled from unlabeled: Exploiting unlabeled data for few-shot deep hdr deghosting. In CVPR, pages 4875–4885, 2021.
- Burst denoising via temporally shifted wavelet transforms. In ECCV, pages 240–256. Springer, 2020.
- Selective transhdr: Transformer-based selective hdr imaging using ghost region mask. In ECCV, pages 288–304. Springer, 2022.
- Deep sr-hdr: Joint learning of super-resolution and high dynamic range imaging for dynamic scenes. IEEE TMM, 25:750–763, 2021.
- Alignment-free hdr deghosting with semantics consistent transformer. In ICCV, 2023.
- Deep learning for hdr imaging: State-of-the-art and future trends. IEEE TPAMI, 44(12):8874–8895, 2021.
- End-to-end learning for image burst deblurring. In ACCV, pages 35–51. Springer, 2017.
- Deep high dynamic range imaging with large foreground motions. In ECCV, pages 117–132, 2018.
- Basis prediction networks for effective burst denoising with large kernels. In CVPR, pages 11844–11853, 2020.
- Attention-guided network for ghost-free high dynamic range imaging. In CVPR, pages 1751–1760, 2019.
- Cycleisp: Real image restoration via improved data synthesis. In CVPR, pages 2696–2705, 2020.
- Multi-stage progressive image restoration. In CVPR, pages 14821–14831, 2021.
- Beyond a gaussian denoiser: Residual learning of deep cnn for image denoising. IEEE TIP, 26(7):3142–3155, 2017.
- Bracketing is all you need: Unifying image restoration and enhancement tasks with multi-exposure images. arXiv preprint arXiv:2401.00766, 2024a.
- NTIRE 2024 challenge on bracketing image restoration and enhancement: Datasets, methods and results. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) Workshops, 2024b.
- Rawhdr: High dynamic range image reconstruction from a single raw image. In ICCV, pages 12334–12344, 2023.