Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Color Image Denoising Using The Green Channel Prior

Zhaoming Kong, Xiaowei Yang

TL;DR

This work introduces GCP-ID, a one-step color image denoising method that exploits the green channel prior (GCP) within a nonlocal transform framework. It guides nonlocal patch search using the green channel, reformulates RGB patches as RGGB arrays, and employs a block circulant representation with a nonlocal t-SVD transform to capture cross-channel correlations and patch redundancy. The approach yields competitive results on real-world color images and videos and extends to multispectral/hyperspectral imaging, demonstrating the utility of spectrum-wise priors in denoising. The findings suggest that incorporating channel-specific priors can enhance denoising performance and may complement deep-learning approaches in mixed pipelines.

Abstract

Noise removal in the standard RGB (sRGB) space remains a challenging task, in that the noise statistics of real-world images can be different in R, G and B channels. In fact, the green channel usually has twice the sampling rate in raw data and a higher signal-to-noise ratio than red/blue ones. However, the green channel prior (GCP) is often understated or ignored in color image denoising since many existing approaches mainly focus on modeling the relationship among image patches. In this paper, we propose a simple and effective one step GCP-based image denoising (GCP-ID) method, which aims to exploit the GCP for denoising in the sRGB space by integrating it into the classic nonlocal transform domain denoising framework. Briefly, we first take advantage of the green channel to guide the search of similar patches, which improves the patch search quality and encourages sparsity in the transform domain. Then we reformulate RGB patches into RGGB arrays to explicitly characterize the density of green samples. The block circulant representation is utilized to capture the cross-channel correlation and the channel redundancy. Experiments on both synthetic and real-world datasets demonstrate the competitive performance of the proposed GCP-ID method for the color image and video denoising tasks. The code is available at github.com/ZhaomingKong/GCP-ID.

Color Image Denoising Using The Green Channel Prior

TL;DR

This work introduces GCP-ID, a one-step color image denoising method that exploits the green channel prior (GCP) within a nonlocal transform framework. It guides nonlocal patch search using the green channel, reformulates RGB patches as RGGB arrays, and employs a block circulant representation with a nonlocal t-SVD transform to capture cross-channel correlations and patch redundancy. The approach yields competitive results on real-world color images and videos and extends to multispectral/hyperspectral imaging, demonstrating the utility of spectrum-wise priors in denoising. The findings suggest that incorporating channel-specific priors can enhance denoising performance and may complement deep-learning approaches in mixed pipelines.

Abstract

Noise removal in the standard RGB (sRGB) space remains a challenging task, in that the noise statistics of real-world images can be different in R, G and B channels. In fact, the green channel usually has twice the sampling rate in raw data and a higher signal-to-noise ratio than red/blue ones. However, the green channel prior (GCP) is often understated or ignored in color image denoising since many existing approaches mainly focus on modeling the relationship among image patches. In this paper, we propose a simple and effective one step GCP-based image denoising (GCP-ID) method, which aims to exploit the GCP for denoising in the sRGB space by integrating it into the classic nonlocal transform domain denoising framework. Briefly, we first take advantage of the green channel to guide the search of similar patches, which improves the patch search quality and encourages sparsity in the transform domain. Then we reformulate RGB patches into RGGB arrays to explicitly characterize the density of green samples. The block circulant representation is utilized to capture the cross-channel correlation and the channel redundancy. Experiments on both synthetic and real-world datasets demonstrate the competitive performance of the proposed GCP-ID method for the color image and video denoising tasks. The code is available at github.com/ZhaomingKong/GCP-ID.
Paper Structure (14 sections, 11 equations, 10 figures, 5 tables, 1 algorithm)

This paper contains 14 sections, 11 equations, 10 figures, 5 tables, 1 algorithm.

Table of Contents

  1. Introduction
  2. Notations and Preliminaries
  3. Related Works
  4. Method
  5. Green Channel Guided Patch Search
  6. RGGB Representation
  7. Nonlocal t-SVD transform
  8. Complexity Analysis
  9. Experiments
  10. Real-world color image denoising
  11. Real-world color video denoising
  12. Extension to other imaging techniques
  13. Discussion
  14. Conclusion

Figures (10)

  • Figure 1: Signal-to-noise ratio (SNR) in RGB channels of 15 real-world images from the CC15 dataset.
  • Figure 2: Flowchart of the proposed GCP-ID method for sRGB color image denoising.
  • Figure 3: Illustration of the Bayer CFA pattern.
  • Figure 4: Block circulant representation for an RGB input.
  • Figure 5: Color image denoising results on the CC15 dataset.
  • ...and 5 more figures