Poisson-Gaussian Noise Reduction Using the Hidden Markov Model in Contourlet Domain for Fluorescence Microscopy...

Area of Science:

• Image processing
• Computational imaging
• Scientific visualization

Background:

• Low-photon imaging in fluorescence microscopy and astronomy yields images with signal-dependent Poisson noise and low signal-to-noise ratios.
• Existing denoising methods often focus on signal-independent Gaussian noise, leaving a gap for complex noise models.

Purpose of the Study:

• To develop an effective denoising algorithm for images affected by combined Poisson-Gaussian noise.
• To improve image quality in low-light scientific imaging applications.

Main Methods:

• Modeling noise as a combination of Poisson and Gaussian distributions.
• Utilizing the contourlet transform for sparse representation of image directional components.
• Applying hidden Markov models to capture spatial and interscale dependencies of transform coefficients.
• Incorporating noise estimation in the transform domain, cycle spinning, and Wiener filtering.

Main Results:

• Demonstrated improved performance of the proposed denoising algorithm through simulations.
• Validated the algorithm's effectiveness on real fluorescence microscopy images.
• Showcased superior noise reduction compared to existing methods for Poisson-Gaussian noise.

Conclusions:

• The proposed algorithm offers a robust solution for denoising images with Poisson-Gaussian noise.
• The integration of contourlet transform and hidden Markov models provides a powerful framework for complex image noise reduction.
• The method significantly enhances image quality in critical scientific imaging domains.