A GPU-accelerated signal processing library.
Project description
This project contains several transformations used for signal processing, along with associated tools.
Transformations
As of v0.2, the included transformations are: discrete Fourier transform (dft), discrete cosine transform (dct), and discrete wavelet transform (dwt).
GPU-Acceleration
Each of these transformations has both a sequential and a GPU-acclerated version. To use the GPU-accelerated version, you would simply import “[transformation_name]Cuda” instead of ‘transformation_name”. For example, you might use
import pygasp.dwtCuda as dwt
instead of using
import pygasp.dwt as dwt
to get the CUDA version of the dwt.
NOTE: To use the CUDA versions, you must have a CUDA-capable GPU and have installed the PyCUDA library. Information on this library can be found at: mathema.tician.de/software/pycuda .
Tools
- As of v0.2, the tools include:
high-, low-, and bandpass filters for the fft
hard and soft thresholding for the dwt
scalogram visualization of the dwt
License
This library is free and open-source software released under the MIT License. The source code is available at https://bitbucket.org/bowmanat/pygasp .
Contact
Please email “bowmanat AT mail DOT gvsu DOT edu” with any questions or comments.
