NECEC 2017 has ended
View analytic
Wednesday, November 15 • 15:20 - 15:40
Towards a Non-local Oriented-Laplacian Decomposition Model

Sign up or log in to save this to your schedule and see who's attending!

The use of non-local means has recently become popular in the computer vision and image processing communities as a method of improving the performance of denoising, as well as other algorithms. Such non-local methods use the entire image instead of just a local neighborhood around each pixel to estimate statistics used in processing the image at a given pixel. Recently, the authors of this work introduced an oriented Laplacian decomposition model in lieu of the usual non-oriented Laplacian decomposition model of Osher, Sole and Vese. Higher quality denoising results were generally obtained with this recently-proposed decomposition model for oriented texture images. Here, a new non-local version of the authors' oriented Laplacian decomposition model is proposed. Experimental results show this new combination of non-local means and Oriented Laplacian decomposition can yield even higher quality denoising results than either non-local means or Oriented Laplacian decomposition separately.

Wednesday November 15, 2017 15:20 - 15:40
Conception Bay North 180 Portugal Cove Road, St. John's, NL, Canada