On the improvement of volterra equation based filtering for image denoising

Eduardo Cuesta, Mokhtar Kirane, Salman A. Malik

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

This paper presents a simple but effective approach for the removal of additive white Gaussian noise from digital images. In our approach, a generalization of a linear heat equation, obtained by replacing time derivative to a fractional time derivative of order between 1 and 2 has been used and a pixel by pixel technique applied. The choice of order of fractional time derivative has been made for each pixel by using structure tensor of image, which allows us to control the diffusion process without introducing nonlinearities in equation as in classical approaches. The proposed model is well posed and numerical scheme adopted is stable. Several experiments showing improvement of our approach visually and in terms of SNR, PSNR are also provided.

Original languageBritish English
Title of host publicationProceedings of the 2011 International Conference on Image Processing, Computer Vision, and Pattern Recognition, IPCV 2011
Pages733-738
Number of pages6
StatePublished - 2011
Event2011 International Conference on Image Processing, Computer Vision, and Pattern Recognition, IPCV 2011 - Las Vegas, NV, United States
Duration: 18 Jul 201121 Jul 2011

Publication series

NameProceedings of the 2011 International Conference on Image Processing, Computer Vision, and Pattern Recognition, IPCV 2011
Volume2

Conference

Conference2011 International Conference on Image Processing, Computer Vision, and Pattern Recognition, IPCV 2011
Country/TerritoryUnited States
CityLas Vegas, NV
Period18/07/1121/07/11

Keywords

  • Convolution quadrature methods
  • Fractional integrals and derivatives
  • Structure tensor
  • Volterra equations

Fingerprint

Dive into the research topics of 'On the improvement of volterra equation based filtering for image denoising'. Together they form a unique fingerprint.

Cite this