Back to Search
Start Over
Infimal convolution of data discrepancies for mixed noise removal
- Publication Year :
- 2016
- Publisher :
- arXiv, 2016.
-
Abstract
- We consider the problem of image denoising in the presence of noise whose statistical properties are a combination of two different distributions. We focus on noise distributions that are frequently considered in applications, in particular mixtures of salt & pepper and Gaussian noise, and Gaussian and Poisson noise. We derive a variational image denoising model that features a total variation regularisation term and a data discrepancy that features the mixed noise as an infimal convolution of discrepancy terms of the single-noise distributions. We give a statistical derivation of this model by joint Maximum A-Posteriori (MAP) estimation, and discuss in particular its interpretation as the MAP of a so-called infinity convolution of two noise distributions. Moreover, classical single-noise models are recovered asymptotically as the weighting parameters go to infinity. The numerical solution of the model is computed using second order Newton-type methods. Numerical results show the decomposition of the noise into its constituting components. The paper is furnished with several numerical experiments and comparisons with other existing methods dealing with the mixed noise case are shown.
- Subjects :
- Applied Mathematics
General Mathematics
010103 numerical & computational mathematics
02 engineering and technology
Total variation denoising
01 natural sciences
Convolution
Gradient noise
Noise
symbols.namesake
Gaussian noise
Optimization and Control (math.OC)
Computer Science::Computer Vision and Pattern Recognition
Statistics
0202 electrical engineering, electronic engineering, information engineering
symbols
FOS: Mathematics
020201 artificial intelligence & image processing
Value noise
0101 mathematics
Image denoising
Focus (optics)
Mathematics - Optimization and Control
Algorithm
Mathematics
Subjects
Details
- Database :
- OpenAIRE
- Accession number :
- edsair.doi.dedup.....f5ff9e5dcda76eeb53ca0d312e71f3d6
- Full Text :
- https://doi.org/10.48550/arxiv.1611.00690