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.