TR2015-106

Accelerated graph-based spectral polynomial filters


    •  Knyazev, A., Malyshev, A., "Accelerated Graph-Based Spectral Polynomial Filters", IEEE International Workshop on Machine Learning for Signal Processing (MLSP), DOI: 10.1109/​MLSP.2015.7324315, September 2015, pp. 1-6.
      BibTeX TR2015-106 PDF
      • @inproceedings{Knyazev2015sep1,
      • author = {Knyazev, A. and Malyshev, A.},
      • title = {Accelerated Graph-Based Spectral Polynomial Filters},
      • booktitle = {IEEE International Workshop on Machine Learning for Signal Processing (MLSP)},
      • year = 2015,
      • pages = {1--6},
      • month = sep,
      • publisher = {IEEE},
      • doi = {10.1109/MLSP.2015.7324315},
      • issn = {1551-2541},
      • url = {https://www.merl.com/publications/TR2015-106}
      • }
  • Research Area:

    Computer Vision

Abstract:

Graph-based spectral denoising is a low-pass filtering using the eigendecomposition of the graph Laplacian matrix of anoisy signal. Polynomial filtering avoids costly computation of the eigendecomposition by projections onto suitable Krylov subspaces. Polynomial filters can be based, e.g.,on the bilateral and guided filters. We propose constructing accelerated polynomial filters by running flexible Krylov subspace based linear and eigenvalue solvers such as the Block Locally Optimal Preconditioned Conjugate Gradient(LOBPCG) method.

 

  • Related News & Events