Unconstrained Structural Similarity-Based Optimization
We establish a general framework, along with a set of algorithms, for the incorporation of the Structural Similarity (SSIM) quality index measure as the fidelity, or “data fitting,” term in objective functions for optimization problems in image processing. The motivation for this approach is to replace the widely used Euclidean distance, known as a poor measure of visual quality, by the SSIM, which has been recognized as one of the best measures of visual closeness. Some experimental results are also presented.
KeywordsVisual Quality Human Visual System Tikhonov Regularization Coordinate Descent Structural Similarity Index Measure
Unable to display preview. Download preview PDF.
- 1.Albuquerque, G., Eisemann, M., Magnor, M.A.: Perception-based visual quality measures. In: 2011 IEEE Conference on Visual Analytics Science and Technology (VAST), pp. 13–20 (2011)Google Scholar
- 2.Brunet, D.: A Study of the Structural Similarity Image Quality Measure with Applications to Image Processing. Ph.D. Thesis, Department of Applied Mathematics, University of Waterloo (2012)Google Scholar
- 6.Boyd, S., Vandenberghe, L.: Convex Optimization. Cambridge University Press (2004)Google Scholar
- 11.Mairal, J., Bach, F., Jenatton, R., Obozinski, G.: Convex Optimization with Sparsity-Inducing Norms. Optimization for Machine Learning. MIT Press (2011)Google Scholar
- 16.Turlach, B.A.: On algorithms for solving least squares problems under an \(\ell _1\) penalty or an \(\ell _1\) constraint. In: Proceedings of the American Statistical Association, Statistical Computing Section, pp. 2572–2577 (2005)Google Scholar