Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Multi-View Learning with Context-Guided Receptance for Image Denoising

Binghong Chen, Tingting Chai, Wei Jiang, Yuanrong Xu, Guanglu Zhou, Xiangqian Wu

TL;DR

This paper tackles real-world image denoising under complex noise with a computationally efficient architecture. It introduces CRWKV, a context-guided receptance framework that fuses Frequency Mix (FMix), Context-guided Token Shift (CTS), and Bidirectional WKV (BiWKV) in a U-Net backbone to enable full pixel-sequence interaction with linear complexity. The approach demonstrates superior quantitative and qualitative performance on real-world datasets (SIDD, ccnoise, PolyU, Urban100GP) and offers strong generalization with reduced inference time and memory usage. The work provides detailed ablations confirming the necessity of FMix and CTS, and positions CRWKV as a practical, high-performance solution for real-world image denoising tasks.

Abstract

Image denoising is essential in low-level vision applications such as photography and automated driving. Existing methods struggle with distinguishing complex noise patterns in real-world scenes and consume significant computational resources due to reliance on Transformer-based models. In this work, the Context-guided Receptance Weighted Key-Value (\M) model is proposed, combining enhanced multi-view feature integration with efficient sequence modeling. Our approach introduces the Context-guided Token Shift (CTS) paradigm, which effectively captures local spatial dependencies and enhance the model's ability to model real-world noise distributions. Additionally, the Frequency Mix (FMix) module extracting frequency-domain features is designed to isolate noise in high-frequency spectra, and is integrated with spatial representations through a multi-view learning process. To improve computational efficiency, the Bidirectional WKV (BiWKV) mechanism is adopted, enabling full pixel-sequence interaction with linear complexity while overcoming the causal selection constraints. The model is validated on multiple real-world image denoising datasets, outperforming the existing state-of-the-art methods quantitatively and reducing inference time up to 40\%. Qualitative results further demonstrate the ability of our model to restore fine details in various scenes.

Multi-View Learning with Context-Guided Receptance for Image Denoising

TL;DR

This paper tackles real-world image denoising under complex noise with a computationally efficient architecture. It introduces CRWKV, a context-guided receptance framework that fuses Frequency Mix (FMix), Context-guided Token Shift (CTS), and Bidirectional WKV (BiWKV) in a U-Net backbone to enable full pixel-sequence interaction with linear complexity. The approach demonstrates superior quantitative and qualitative performance on real-world datasets (SIDD, ccnoise, PolyU, Urban100GP) and offers strong generalization with reduced inference time and memory usage. The work provides detailed ablations confirming the necessity of FMix and CTS, and positions CRWKV as a practical, high-performance solution for real-world image denoising tasks.

Abstract

Image denoising is essential in low-level vision applications such as photography and automated driving. Existing methods struggle with distinguishing complex noise patterns in real-world scenes and consume significant computational resources due to reliance on Transformer-based models. In this work, the Context-guided Receptance Weighted Key-Value (\M) model is proposed, combining enhanced multi-view feature integration with efficient sequence modeling. Our approach introduces the Context-guided Token Shift (CTS) paradigm, which effectively captures local spatial dependencies and enhance the model's ability to model real-world noise distributions. Additionally, the Frequency Mix (FMix) module extracting frequency-domain features is designed to isolate noise in high-frequency spectra, and is integrated with spatial representations through a multi-view learning process. To improve computational efficiency, the Bidirectional WKV (BiWKV) mechanism is adopted, enabling full pixel-sequence interaction with linear complexity while overcoming the causal selection constraints. The model is validated on multiple real-world image denoising datasets, outperforming the existing state-of-the-art methods quantitatively and reducing inference time up to 40\%. Qualitative results further demonstrate the ability of our model to restore fine details in various scenes.
Paper Structure (25 sections, 11 equations, 7 figures, 7 tables, 1 algorithm)

This paper contains 25 sections, 11 equations, 7 figures, 7 tables, 1 algorithm.

Table of Contents

  1. Introduction
  2. Related Works
  3. Real-world Image Denoising
  4. RWKV Models
  5. Methodology
  6. Overall Architecture
  7. Context Receptance Module
  8. Context-guided Token Shift.
  9. Bidirectional WKV operation.
  10. Frequency Mix Module
  11. Loss Function
  12. Experiments
  13. Experiment Setup
  14. Datasets and metrics.
  15. Implementation details.
  16. ...and 10 more sections

Figures (7)

  • Figure 1: Comparison of existing image denoising methods. (a) Visual results on Urban100 dataset: our method preserves fine details under complex noise pattern, while others suffer from information losses or artifacts. (b) Computational complexity across input scales for Transformer- and state space-based models.
  • Figure 2: Architecture of the proposed CRWKV model, and (a) Frequency Mix Module (FMix), (b) Context Receptance Module (CRM) and (c) Context-guided Token Shift (CTS).
  • Figure 3: Illustration of different token shift mechanisms.
  • Figure 4: Selected visual results on ccnoise and PolyU dataset.
  • Figure 5: Selected visual results on SIDD dataset.
  • ...and 2 more figures