Non-local total variation regularization approach for image restoration under a Poisson degradation

ABSTRACT

Poisson noise (also known as shot or photon noise) arises due to the lack of information during the image acquisition phase, it is quite common in the field of microscopic or astronomical imaging applications. In this paper, we propose a non-local total variation regularization framework with a p-norm driven data-fidelity for denoising the Poissonian images. In precise, the energy functional is derived using a Maximum A Posteriori estimator of the Poisson probability density function. The diffusion amounts to a non-local total variation minimization process, which eventually preserves fine structures while denoising the data. The numerical solution is sought under a fast converging split-Bregman iterative scheme. The proposed model is compared visually and statistically with the state-of-the-art Poisson denoising models.

KEYWORDS:

• Image restoration
• Poisson noise
• non-local total variation
• split-Bregman iteration

Acknowledgments

Mr. Shivarama Holla would like to thank the Ministry of Human Resource Development, Government of India, for providing the financial assistance to pursue Ph.D. research work at National Institute of Technology, Karnataka, India.

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Funding

Dr. Jidesh would like to thank the Department of Science and Technology (Science and Engineering Research Board), Government of India for providing the financial support under the Project Grant No. ECR/2017/000230.