RASLF: Representation-Aware State Space Model for Light Field Super-Resolution

2026-03-17

AI Summary

This work addresses the limitations of existing state-space-model-based light field super-resolution methods, which often fail to effectively integrate complementary information across multiple representations, leading to loss of texture details and misalignment in view geometry. To overcome these challenges, we propose a representation-aware state space framework that explicitly models structural relationships among diverse light field representations—such as epipolar plane images and sub-aperture views—to enable accurate and efficient super-resolution reconstruction. Key innovations include a progressive geometric refinement module for view alignment correction, a representation-aware asymmetric scanning mechanism tailored to heterogeneous representation characteristics, and a dual-anchor hierarchical feature aggregation strategy. Extensive experiments demonstrate that our method achieves state-of-the-art reconstruction performance on multiple public benchmarks while maintaining superior computational efficiency.

Technology Category

• Computer Vision: Representation Learning for Vision
• Machine Learning: Representation Learning
• Intelligent Robots: State Estimation

Application Category

• Search and Retrieval-Augmented AI: Web query analysis, representation and understanding
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs

Abstract

Current SSM-based light field super-resolution (LFSR) methods often fail to fully leverage the complementarity among various LF representations, leading to the loss of fine textures and geometric misalignments across views. To address these issues, we propose RASLF, a representation-aware state-space framework that explicitly models structural correlations across multiple LF representations. Specifically, a Progressive Geometric Refinement (PGR) block is created that uses a panoramic epipolar representation to explicitly encode multi-view parallax differences, thereby enabling integration across different LF representations. Furthermore, we introduce a Representation Aware Asymmetric Scanning (RAAS) mechanism that dynamically adjusts scanning paths based on the physical properties of different representation spaces, optimizing the balance between performance and efficiency through path pruning. Additionally, a Dual-Anchor Aggregation (DAA) module improves hierarchical feature flow, reducing redundant deeplayer features and prioritizing important reconstruction information. Experiments on various public benchmarks show that RASLF achieves the highest reconstruction accuracy while remaining highly computationally efficient.

Problem

• light field super-resolution
• state space model
• representation complementarity
• geometric misalignment
• texture loss

Innovation

• State Space Model
• Light Field Super-Resolution
• Representation-Aware Scanning
• Geometric Refinement
• Feature Aggregation