Loop-Mamba: A Loop Mamba with Degradation-Aware and Shared Memory for Old Photo Restoration

2026-08-03

AI Summary

This work addresses the challenging problem of restoring old photographs, which often suffer from multiple coupled degradations—including scratches, cracks, fading, blur, noise, and missing regions—that severely compromise both visual quality and semantic content. The authors propose Loop-Mamba, a lightweight recurrent state-space model that formulates image restoration as a continuously evolving recovery process. Key innovations include a semantics-guided degradation estimator for explicitly modeling heterogeneous degradations, a shared structural memory Mamba module enabling long-range state evolution across iterations, and an efficient multi-directional scanning strategy. Evaluated on the SynOld benchmark, Loop-Mamba outperforms state-of-the-art methods in both conventional metrics and the newly introduced task-oriented ODRS assessment, achieving substantial improvements in restoration quality and structural fidelity.

Abstract

Old photographs often suffer from multiple coupled degradations, including scratches, cracks, fading, blur, noise, and missing regions, severely degrading both visual quality and semantic content. We propose Loop-Mamba, a lightweight loop-based state-space framework that formulates old photo restoration as progressive state evolution, where a persis- tent restoration state is continuously propagated and refined through iterative computation. Specifically, we introduce a Semantic-Guided Degradation Estimator (SGDE) to explicitly model heterogeneous degradations by jointly predicting local degradation maps and global degradation scores, providing degradation-aware guidance for state evolution. We further develop a Shared Structural Memory Mamba (S$^2$M- Mamba), which maintains a persistent restoration state across iterations, enabling persistent state evolution through shared structural memory for robust long-range structural reconstruction. Benefiting from first-order state recursion, Loop-Mamba propagates latent restoration states through recurrent tran- sitions instead of repeatedly stacking deep feature transformations, thereby alleviating gradient dilution while avoiding the computational overhead inherent in iterative CNN- and Transformer-based restoration frameworks. A lightweight multi-directional scanning strategy further enhances direc- tional information aggregation and preserves structural continuity. To better evaluate restoration quality, we introduce the task-oriented Old Photo Damage Recovery Score (ODRS), which jointly measures degradation recovery and structural reconstruction fidelity. Experimental results on the public SynOld benchmark demonstrate that Loop-Mamba consistently outperforms previous state-of-the-art methods across both conventional restoration metrics and the proposed ODRS.

Problem

• old photo restoration
• image degradation
• structural reconstruction
• degradation recovery
• visual quality

Innovation

• Loop-Mamba
• Degradation-Aware Estimation
• Shared Structural Memory
• State-Space Model
• Old Photo Restoration