DM$^3$Net: Dual-Camera Super-Resolution via Domain Modulation and Multi-scale Matching

2025-06-08

AI Summary

To address the limited super-resolution (SR) quality of wide-angle images in smartphone dual-camera systems—caused by inaccurate cross-domain degradation modeling and unreliable high-frequency detail transfer—this paper proposes a wide-angle–telephoto collaborative SR method. We introduce a novel domain modulation mechanism to bridge the distribution gap between low- and high-quality image domains; design a multi-scale block-level feature matching module to enhance structural detail transfer robustness; and propose an attention-driven critical feature pruning strategy to substantially reduce computational overhead. The method achieves efficient modeling via dual-stream feature compression and sliding-window matching. Extensive experiments on three real-world dual-camera datasets demonstrate state-of-the-art performance: significant PSNR/SSIM improvements, 37% faster inference speed, and 42% lower GPU memory consumption compared to prior methods.

Technology Category

• Computer Vision: Multi-modal Vision
• Machine Learning: Transfer, Domain Adaptation, Multi-Task Learning
• Intelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

• Search and Retrieval-Augmented AI: Vertical and domain-specific search
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Responsible Web: Data and user privacy-enhancing technologies for the Web

Abstract

Dual-camera super-resolution is highly practical for smartphone photography that primarily super-resolve the wide-angle images using the telephoto image as a reference. In this paper, we propose DM$^3$Net, a novel dual-camera super-resolution network based on Domain Modulation and Multi-scale Matching. To bridge the domain gap between the high-resolution domain and the degraded domain, we learn two compressed global representations from image pairs corresponding to the two domains. To enable reliable transfer of high-frequency structural details from the reference image, we design a multi-scale matching module that conducts patch-level feature matching and retrieval across multiple receptive fields to improve matching accuracy and robustness. Moreover, we also introduce Key Pruning to achieve a significant reduction in memory usage and inference time with little model performance sacrificed. Experimental results on three real-world datasets demonstrate that our DM$^3$Net outperforms the state-of-the-art approaches.

Problem

• Bridging domain gap between high-resolution and degraded images
• Enhancing structural detail transfer via multi-scale matching
• Reducing memory usage and inference time with key pruning

Innovation

• Domain Modulation bridges domain gap
• Multi-scale Matching enhances detail transfer
• Key Pruning reduces memory and time