Conference paperBAŘINA David, NAJMAN Pavel, KLEPÁRNÍK Petr, KULA Michal and ZEMČÍK Pavel. The Parallel Algorithm for the 2D Discrete Wavelet Transform. In: Ninth International Conference on Graphic and Image Processing (ICGIP 2017). Qingdao: SPIE  the international society for optics and photonics, 2017, pp. 16. ISBN 9781510617414. ISSN 0277786X.  Publication language:  english 

Original title:  The Parallel Algorithm for the 2D Discrete Wavelet Transform 

Title (cs):  Paralelní algoritmus pro 2D diskrétní vlnkovou transformaci 

Pages:  16 

Proceedings:  Ninth International Conference on Graphic and Image Processing (ICGIP 2017) 

Conference:  International Conference on Graphic and Image Processing 2017 

Place:  Qingdao, CN 

Year:  2017 

ISBN:  9781510617414 

Journal:  Proceedings of SPIE, Vol. 10615, No. 4, CZ 

ISSN:  0277786X 

DOI:  10.1117/12.2302881 

Publisher:  SPIE  the international society for optics and photonics 

Files:  

 Keywords 

discrete wavelet transform, lifting scheme, multicore processors, parallel architecture 
Annotation 

The discrete wavelet transform can be found at the heart of many imageprocessing algorithms. Until now, the transform on generalpurpose processors (CPUs) was mostly computed using a separable lifting scheme. As the lifting scheme consists of a small number of operations, it is preferred for processing using singlecore CPUs. However, considering a parallel processing using multicore processors, this scheme is inappropriate due to a large number of steps. On such architectures, the number of steps corresponds to the number of points that represent the exchange of data. Consequently, these points often form a performance bottleneck. Our approach appropriately rearranges calculations inside the transform, and thereby reduces the number of steps. In other words, we propose a new scheme that is friendly to parallel environments. When evaluating on multicore CPUs, we consistently overcome the original lifting scheme. The evaluation was performed on 61core Intel Xeon Phi and 8core Intel Xeon processors. 
Abstract 

The discrete wavelet transform can be found at the heart of many imageprocessing algorithms.
Until now, the transform on generalpurpose processors (CPUs) was mostly computed using a separable lifting scheme.
As the lifting scheme consists of a small number of operations, it is preferred for processing using singlecore CPUs.
However, considering a parallel processing using multicore processors, this scheme is inappropriate due to a large number of steps.
On such architectures, the number of steps corresponds to the number of points that represent the exchange of data.
Consequently, these points often form a performance bottleneck.
Our approach appropriately rearranges calculations inside the transform, and thereby reduces the number of steps.
In other words, we propose a new scheme that is friendly to parallel environments.
When evaluating on multicore CPUs, we consistently overcome the original lifting scheme.
The evaluation was performed on 61core Intel Xeon Phi and 8core Intel Xeon processors. 
BibTeX: 

@INPROCEEDINGS{
author = {David Ba{\v{r}}ina and Pavel Najman and Petr
Klep{\'{a}}rn{\'{i}}k and Michal Kula and Pavel
Zem{\v{c}}{\'{i}}k},
title = {The Parallel Algorithm for the 2D Discrete
Wavelet Transform},
pages = {16},
booktitle = {Ninth International Conference on Graphic and Image
Processing (ICGIP 2017)},
journal = {Proceedings of SPIE},
volume = 10615,
number = 4,
year = 2017,
location = {Qingdao, CN},
publisher = {SPIE  the international society for optics and photonics},
ISBN = {9781510617414},
ISSN = {0277786X},
doi = {10.1117/12.2302881},
language = {english},
url = {http://www.fit.vutbr.cz/research/view_pub.php.en.iso88592?id=11530}
} 
