Skip to content
master
Switch branches/tags
Code

Latest commit

 

Git stats

Files

Permalink
Failed to load latest commit information.
Type
Name
Latest commit message
Commit time
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

tfShearlab

Tensorflow implementation of Shearlab, including a python wrapper of the Julia Shearlab APi

Installation and dependencies

tfShearlab can be easily installed by running

$ pip install https://github.com/arsenal9971/tfshearlab/archive/master.zip

This package has the next dependencies

  • Julia language: One can either precompiled packages or build from source. This package requires the Julia version 0.6 or higher.

  • Shearlab.jl: To install the library in Julia 0.6.x one needs to run the command julia -e 'Pkg.add("Shearlab").

  • Pyjulia: One can install the Python API of Julia with the command pip install julia, for more details on installation check the documentation.

    • One also needs to make the Julia and Python enviroment to coincide running the command julia -e 'ENV["PYTHON"]="<your-python-executable>"; Pkg.add("PyCall"); Pkg.build("PyCall")'. One can find its python executable path by running on the terminal $(which python).
  • SSL certificates: Sometimes you need to give (and add to bashrc.) the SSL certificates path using export SSL_CERT_FILE=/etc/ssl/ca-bundle.pem.

Description

Shearlab.jl is a Julia Library with toolbox for two- and threedimensional data processing using the Shearlet system as basis functions which generates an sparse representation of cartoon-like functions.

tfShearlab is a tensorflow implementation of the Shearlet transform using the Julia API as backend. The reason for this implementation lies in mainly in the GPU-functionalities of tensorflow that accelerates the fft-based convolutions; in comparison with the version without tensorflow, the Shearlet decomposition and recosntruction are about 30x faster in a GTX 1080 graphic card.

This package also contains a python wrapper of the Julia API, so one can perform the Shearlet transform without tensorflow.

For the 2D version one has three important functions:

  • Generate the Shearlet System.
getshearletsystem2D(rows,cols,nScales,shearLevels,full,directionalFilter,scalingFilter);
  • Decoding of a signal X.
tfsheardec2D(Xtf, tfshearlets)  
  • Reconstruction of a signal Xtf.
tfshearrec2D(coeffstf,tfshearlets,tfdualFrameWeights )

For more detailed usage functionalities check the original Shearlab manual, or examples for scientific reference one can also read "ShearLab 3D: Faithful Digital Shearlet Transforms Based on Compactly Supported Shearlets".

About

Tensorflow implementation of Shearlab, including a python wrapper of the Julia Shearlab APi

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages