Detail publikace

Segmentwise Computation of 2D Forward Discrete Wavelet Transform

Originální název

Segmentwise Computation of 2D Forward Discrete Wavelet Transform

Anglický název

Segmentwise Computation of 2D Forward Discrete Wavelet Transform

Jazyk

en

Originální abstrakt

In this paper we present a new algorithm for segmentwise computation of 2D forward discrete-time wavelet transform. The idea of the algorithm is that the 2D discrete signal (image) is divided into rectangular segments and the computation of the wavelet transform coefficients can be performed on each of the segments independently. The algorithm introduces substantial advantage in that the transform can be performed on segments in parallel with the cost of a slight redundancy (several coefficients are computed more than once). The size of the segments is arbitrary in both the vertical and the horizontal directions, so there is place for performance tuning. We also present our implementation of the algorithm in parallel 2D forward DTWT using C++ and Intel Threading Building Blocks library.

Anglický abstrakt

In this paper we present a new algorithm for segmentwise computation of 2D forward discrete-time wavelet transform. The idea of the algorithm is that the 2D discrete signal (image) is divided into rectangular segments and the computation of the wavelet transform coefficients can be performed on each of the segments independently. The algorithm introduces substantial advantage in that the transform can be performed on segments in parallel with the cost of a slight redundancy (several coefficients are computed more than once). The size of the segments is arbitrary in both the vertical and the horizontal directions, so there is place for performance tuning. We also present our implementation of the algorithm in parallel 2D forward DTWT using C++ and Intel Threading Building Blocks library.

Dokumenty

BibTex


@inproceedings{BUT35299,
  author="Zdeněk {Průša} and Pavel {Rajmic} and Jan {Malý}",
  title="Segmentwise Computation of 2D Forward Discrete Wavelet Transform",
  annote="In this paper we present a new algorithm for segmentwise computation of 2D forward discrete-time wavelet transform. The idea of the algorithm is that the 2D discrete signal (image) is divided into rectangular segments and the computation of the wavelet transform coefficients can be performed on each of the segments independently. The algorithm introduces substantial advantage in that the transform can be performed on segments in parallel with the cost of a slight redundancy (several coefficients are computed more than once). The size of the segments is arbitrary in both the vertical and the horizontal directions, so there is place for performance tuning. We also present our implementation of the algorithm in parallel 2D forward DTWT using C++ and Intel Threading Building Blocks library.",
  address="ASSZISZTENCIA Congress Bureau",
  booktitle="33rd International Conference on TELECOMMUNICATIONS AND SIGNAL PROCESSING",
  chapter="35299",
  howpublished="online",
  institution="ASSZISZTENCIA Congress Bureau",
  year="2010",
  month="august",
  pages="1--4",
  publisher="ASSZISZTENCIA Congress Bureau",
  type="conference paper"
}