Real FFT based on graphics processing unit for image processing

doi:null

您当前的位置：

首页 >

文章列表页 >

Real FFT based on graphics processing unit for image processing

Article | 更新时间：2020-08-13

- Real FFT based on graphics processing unit for image processing
- Optics and Precision Engineering Vol. 16, Issue 12, Pages: 2414-2420(2008)
- 作者机构：
  
  1. 中国科学院研究生院
  2. 长春光学精密机械与物理研究所
- 作者简介：
- 基金信息：
- DOI：
  CLC：
- Received：06 August 2008，
  
  Revised：07 October 2008，
  
  Published Online：25 December 2008，
  
  Published：25 December 2008
- 稿件说明：
移动端阅览
李仕. 基于图形处理器的实数FFT在图像处理上的应用[J]. 光学精密工程, 2008,16(12):2414-2420

Real FFT based on graphics processing unit for image processing[J]. Optics and precision engineering, 2008, 16(12): 2414-2420.
李仕. 基于图形处理器的实数FFT在图像处理上的应用[J]. 光学精密工程, 2008,16(12):2414-2420 DOI：

Real FFT based on graphics processing unit for image processing[J]. Optics and precision engineering, 2008, 16(12): 2414-2420. DOI：

摘要

对常用的实数FFT算法进行改进，在此基础上提出基于图形处理器(GPU)平台的实数FFT算法。常用的实数FFT算法降低传统复数FFT的运算量约40%，文中实数FFT算法的运算量能减少到复数FFT的50%。通过对存储数据的特殊配置，改进后的实数FFT算法能由一维拓展到二维。并通过调用CUFFT库的一维复数FFT函数，实现实数FFT算法的并行运算。实验中通过维纳滤波算法的应用，对实数FFT的精确度及效率进行检验。图像的处理效果表明文中实数FFT算法是正确、可行的。文中实数FFT算法可以帮助维纳滤波算法在19.26ms内恢复一帧2048x2048灰度模糊图像，该速度是在GPU上使用复数FFT的维纳滤波的2.34倍，是在CPU上使用实数FFT的维纳滤波的37.46倍。实验表明文中算法具有一定的实用价值。

Abstract

Through the analysis of the general real Fast Fourier Transform (FFT) algorithm

a improved real FFT algorithm is presented based on Graphics Processing Unit（GPU）. General real FFT reduces the computation consumption of the conventional complex FFT by about 40%

and the presented real FFT improves the reduction to 50%. On the basis of the specific configuration of the data storage

the presented real FFT can be applied in two directions continuously to calculate the two dimensional Fourier transform. What’s more

the parallel real FFT transform can be implemented by using the 1D complex FFT function from the CUFFT library. The precision and the efficiency of the presented real FFT are inspected by the Wiener filter’s application. The image processing results demonstrate that the presented real FFT is operating properly and working well. With the help of the presented real FFT

Wiener filter can restore a 2048x2048 8-bit image degraded by motion in 19.26ms. It is 2.34X speed compared to the Wiener filter based on complex FFT on GPU

and is 37.46X speed compared to the Wiener filter based on real FFT on CPU. The experiment results show the presented algorithm is of practical use.

关键词

Keywords

references

Views

261

下载量

CSCD

Alert me when the article has been cited

提交

Tools

Publicity Resources

Hyperspectral image restoration via weighted Schatten norm low-rank representation

Astronomical image restoration basedMaximum-likelihood incorporated parameter estimation

Image restoration based on real time wave-front information

Parallelity improvement of object function for phase diversity

Image restoration by phase-diverse speckle

Related Author

Xiao-zhen XIE

Qian-ying ZHANG

SHEN Chen

WEI Xiao-feng

GENG Ze-xun

MA Xin-xue

WU Yuan-hao

ZHAO Jin-yu

Related Institution

College of Science, Noutheast A & F University, Yangling

Shenzhen Tourism College, Jinan University

Institute of Geographic Space Information, Information Engineering University

Graduate University of Chinese Academy of Sciences

Changchun Institute of Optics, Fine Mechanics and Physics, Chinese Academy of Sciences

AI问答

Address：No.3888 Dong Nanhu Road, Changchun, Jilin, China Postal code：130033
Tel：0431-86176855 Email：gxjmgc@ciomp.ac.cn
Technical support is provided by Beijing Founder electronics co., LTD 吉ICP备11002662号-17 京公网安备11010802024621
It is recommended to read the content of this site in Chrome&IE9+. Please switch to extreme mode in browser 360.
Cookies We use cookies to help provide and enhance our service and tailor content. By continuing, you agree to the use of cookies.

⁰