Using Partial Differential Equations to Filter Images with High Noise Levels

Using Partial Differential Equations to Filter Images with High Noise Levels

Paper in proceedings (conference paper)

The paper discusses image noise suppression methods based on the solution of partial differential equations. The individual techniques are applied to a set of test images and then compared with respect to the obtained noise suppression levels. One of the presented procedures is image denoising via the heat conduction formula, which constitutes an equivalent of the Gaussian filter. Further, a bilateral filter is derived; this approach better preserves the edges of objects within the examined image. The last technique analyzed is anisotropic diffusion (Perona-Malik). Importantly, the signal-to-noise ratio values obtained with these different filtering methods are compared in the final section of the article.

The paper discusses image noise suppression methods based on the solution of partial differential equations. The individual techniques are applied to a set of test images and then compared with respect to the obtained noise suppression levels. One of the presented procedures is image denoising via the heat conduction formula, which constitutes an equivalent of the Gaussian filter. Further, a bilateral filter is derived; this approach better preserves the edges of objects within the examined image. The last technique analyzed is anisotropic diffusion (Perona-Malik). Importantly, the signal-to-noise ratio values obtained with these different filtering methods are compared in the final section of the article.

Noise filtering, Heat equation, Perona-Malik, Image processing

Noise filtering, Heat equation, Perona-Malik, Image processing

SLIŽ, J.

2017

12.09.2016

VUT FEKT UTEE

Brno

9788021453876

International Interdisciplinary PhD Workshop 2016 Proceedings

82

86

5

http://www.utee.feec.vutbr.cz/iiphdw2016/