loading page

Accelerating the Gabor Transform with a GPU for SAR Image Compression
  • Conner McInnes,
  • Shadi Alawneh
Conner McInnes
Oakland University

Corresponding Author:cmcinnes@oakland.edu

Author Profile
Shadi Alawneh
Oakland University
Author Profile

Abstract

The Gabor transform can be utilized in an algorithm for compression due to its ability to allow the user to isolate high frequency information to filter. This transform can be implemented using FFT’s to aid in calculating the Gabor coefficients of a particular image. In the C++ programming language, an open source library exists called FFTW that is able to perform FFT’s quickly on the CPU. cuFFT does have a bottleneck during the initial allocation of the input, output, and plan for the desired FFT, but with larger images these became less and less impactful. Image compression algorithms using the Gabor transform can benefit in reduced computational time from cuFFT’s functions.