convolutional sparse coding codes

This library provides functions for sparse coding and dictionary learning, together with miscellaneous support functions for signal and image processing with sparse representations. The sparse coding algorithms are based on the ADMM framework; while similar codes for some of these functions can be found elsewhere, those provided here include enhancements that are not present in other publicly available codes.

There are two versions of SPORCO, one implemented in Matlab, and the other in Python; the language is indicated in the version number by an 'M' or 'P' respectively.

Matlab Version

The Matlab version is available from the download link at the bottom of this page. Please read the instructions in the ReadMe file in the distribution before using the library.

Python Version

The Python version is available on GitHub and on PyPI. The current release can be installed using the source package available from PyPI or by doing "pip install sporco" (assuming that pip is installed and the command is issued with root privileges). Documentation for the most recent release is available at PythonHosted, and documentation for the current version on GitHub is available at ReadTheDocs.

The most significant contribution of this library is the inclusion of implementations of recent efficient algorithms for convolutional sparse coding and dictionary learning. These algorithms are described in the papers

• Brendt Wohlberg, " Efficient Convolutional Sparse Coding", in  Proceedings of IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP), (Florence, Italy), pp. 7173-7177, May 2014
• Brendt Wohlberg, " Efficient Algorithms for Convolutional Sparse Representations",  IEEE Transactions on Image Processing, vol. 25, no. 1, pp. 301-315, Jan 2016
• Brendt Wohlberg, " Convolutional Sparse Representation of Color Images", in  Proceedings of the IEEE Southwest Symposium on Image Analysis and Interpretation (SSIAI), (Santa Fe, NM, USA), Mar 2016
• Brendt Wohlberg, " Convolutional Sparse Representations as an Image Model for Impulse Noise Restoration", in  Proceedings of the IEEE Image, Video, and Multidimensional Signal Processing Workshop (IVMSP), (Bordeaux, France), Jul 2016
• Brendt Wohlberg, " Boundary Handling for Convolutional Sparse Representations", in  Proceedings of IEEE International Conference on Image Processing (ICIP), (Phoenix, USA), Sep 2016

This library has been used in a variety of signal and image processing applications of convolutional sparse coding methods, including

• Brendt Wohlberg, " Endogenous Convolutional Sparse Representations for Translation Invariant Image Subspace Models", in  Proceedings of IEEE International Conference on Image Processing (ICIP), (Paris, France), pp. 2859-2863, Oct 2014
• Ping-Keng Jao, Yi-Hsuan Yang and Brendt Wohlberg, " Informed Monaural Source Separation of Music based on Convolutional Sparse Coding", in  Proceedings of IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP), (Brisbane, Australia), pp. 236-240, Apr 2015
• Diego Carrera, Giacomo Boracchi, Alessandro Foi and Brendt Wohlberg, " Detecting Anomalous Structures by Convolutional Sparse Models", in  The International Joint Conference on Neural Networks (IJCNN), (Killarney, Ireland), pp. 1--8, Jul 2015
• Andrea Cogliati, Zhiyao Duan and Brendt Wohlberg, " Piano Music Transcription with Fast Convolutional Sparse Coding", in  Proceedings of the IEEE International Workshop on Machine Learning for Signal Processing, (Boston, MA, USA), pp. 1-6, Sep 2015
• Ping-Keng Jao, Li Su, Yi-Hsuan Yang and Brendt Wohlberg, " Monaural Music Source Separation using Convolutional Sparse Coding",  IEEE/ACM Transactions on Audio, Speech, and Language Processing, 2016
• Andrea Cogliati, Zhiyao Duan and Brendt Wohlberg, " Context-dependent Piano Music Transcription with Convolutional Sparse Coding",  IEEE/ACM Transactions on Audio, Speech, and Language Processing, 2016

License

This library was developed at Los Alamos National Laboratory, and has been approved for public release under the approval number LA-CC-14-057. It is made available under the terms of the BSD 3-Clause License.

Download

Download sporco-m0.0.5.tar.gz (gzipped tar file, 620k)