# A Benchmark for Heterogeneous Stereo Deblurring with Physically- and Epipolar-constrained Cross Attention

arXiv: https://arxiv.org/abs/2606.25962 · HTML: https://mlsanity.com/paper/2606.25962/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 23 of its 35 references that are in the map.
- First posted on arXiv 24 Jun 2026.

## Builds on (most cited first, 23 of 23)

- [Restormer: Efficient Transformer for High-Resolution Image Restoration](https://mlsanity.com/paper/2111.09881/) (2021)
- [Accurate, Large Minibatch SGD: Training ImageNet in 1 Hour](https://mlsanity.com/paper/1706.02677/) (2017)
- [Deep Multi-scale Convolutional Neural Network for Dynamic Scene Deblurring](https://mlsanity.com/paper/1612.02177/) (2016)
- [Simple Baselines for Image Restoration](https://mlsanity.com/paper/2204.04676/) (2022)
- [Rethinking Coarse-to-Fine Approach in Single Image Deblurring](https://mlsanity.com/paper/2108.05054/) (2021)
- [Human-Aware Motion Deblurring](https://mlsanity.com/paper/2001.06816/) (2020)
- [Efficient Frequency Domain-based Transformers for High-Quality Image Deblurring](https://mlsanity.com/paper/2211.12250/) (2022)
- [NAFSSR: Stereo Image Super-Resolution Using NAFNet](https://mlsanity.com/paper/2204.08714/) (2022)
- [Mobile Computational Photography: A Tour](https://mlsanity.com/paper/2102.09000/) (2021)
- [DAVANet: Stereo Deblurring with View Aggregation](https://mlsanity.com/paper/1904.05065/) (2019)
- [Symmetric Parallax Attention for Stereo Image Super-Resolution](https://mlsanity.com/paper/2011.03802/) (2020)
- [Dual-Camera Super-Resolution with Aligned Attention Modules](https://mlsanity.com/paper/2109.01349/) (2021)
- [Reference-based Video Super-Resolution Using Multi-Camera Video Triplets](https://mlsanity.com/paper/2203.14537/) (2022)
- [Joint Stereo Video Deblurring, Scene Flow Estimation and Moving Object Segmentation](https://mlsanity.com/paper/1910.02442/) (2019)
- [Stereo Video Deblurring](https://mlsanity.com/paper/1607.08421/) (2016)
- [Simultaneous Stereo Video Deblurring and Scene Flow Estimation](https://mlsanity.com/paper/1704.03273/) (2017)
- [Dual-Camera Joint Deblurring-Denoising](https://mlsanity.com/paper/2309.08826/) (2023)
- [Cross-View Hierarchy Network for Stereo Image Super-Resolution](https://mlsanity.com/paper/2304.06236/) (2023)
- [Learning Parallax for Stereo Event-based Motion Deblurring](https://mlsanity.com/paper/2309.09513/) (2023)
- [KeDuSR: Real-World Dual-Lens Super-Resolution via Kernel-Free Matching](https://mlsanity.com/paper/2312.17050/) (2023)
- [XYScanNet: A State Space Model for Single Image Deblurring](https://mlsanity.com/paper/2412.10338/) (2024)
- [RefVSR++: Exploiting Reference Inputs for Reference-based Video Super-resolution](https://mlsanity.com/paper/2307.02897/) (2023)
- [Deep Hybrid Camera Deblurring for Smartphone Cameras](https://mlsanity.com/paper/2312.13317/) (2023)

## Abstract (arXiv)

Modern stereo-capable smartphones enable immersive XR content capture. However, hardware heterogeneity across camera modules often causes severe asymmetric blur artifacts. Existing methods and benchmarks largely assume homogeneous stereo setups and therefore do not explicitly address such asymmetric degradation. To bridge this gap, we present a dedicated framework for heterogeneous stereo deblurring. First, we introduce the heterogeneous stereo deblurring (HSD) dataset, constructed from real smartphone stereo captures via multi-frame integration. Second, we propose physically- and epipolar-constrained cross attention (PECA), a lightweight module that restricts cross-view matching to an epipolar search window bounded by a optics-derived disparity upper bound. By enforcing physically valid disparity constraints, PECA enables efficient and reliable cross-view feature fusion. Moreover, our confidence-weighted attention with residual fusion emphasizes cross-guided deblurring when correspondences are reliable, while naturally falling back to self-deblurring in occluded or unreliable regions. PECA is architecture-agnostic and consistently improves CNN-, Transformer-, and NAFNet-based baselines. Extensive experiments on HSD show that PECA-enhanced models achieve improved restoration performance with favorable efficiency.