HIMG

HIMG (Hermes for Images) is an experimental C++ image compression tool based on adaptive hp-FEM and the Hermes2D library. It is distributed under the GNU General Public License.

Download and Installation

Detailed installation instructions for Linux and Windows MSVC can be found in the User documentation.

Support and User Community

HIMG is an advanced application of Hermes2D and thus use the Hermes2D mailing list if you have questions or need help. Use IRC channel irc.freenode.net, room #hermes2d to chat about development in real-time.

Development

HIMG is developed by an open source community around the hp-FEM group at the University of Nevada, Reno (UNR). You are very welcome to join our development team and contribute in any way. Read here How to Submit the First Patch and submit your first patch. Doxygen documentation for HIMG is in progress. Our further development plans are listed below.

Development Plans

• The most basic operation in HIMG is projecting the image onto a finite element mesh. Currently we use an orthogonal projection which requires numerical integration of products of FE basis functions with the image. This is very slow since the image is defined pixel-wise. Speeding up the projection process is our most important goal. Here are some ideas related to this:
  • Replace UMFpack with faster iterative matrix solvers for the projection matrices.
  • Implement local projections which are more efficient than the current global projections.
  • Study the influence of the projection type on the behavior of the image compression algorithm.
• Perform tests with discontinuous (L2) approximations. This might yield better results for images with large local variations. Hermes2D does L2 adaptivity by default, so this should be quite simple.
• Design an efficient data format to store compressed images and their meshes.
• Perform a systematic performance comparison with standard image compression techniques such as JPEG 2000.
• Extend HIMG to movies compression.

Contributors

* Pavel Solin (project leader, University of Nevada, Reno)
* David Andrs (formerly University of Texas at El Paso)
* Ivo Hanak (formerly University of Nevada, Reno)
For more details see the AUTHORS file.

Sample Images

Satellite image of Earth, compressed with adaptive hp-FEM