This module encapsulates algorithms for the restoration of images and it specializes on retinal images. The following names are used to distinguish projects and algorithms separately:
- RRtoolbox is a python package which contains source code designed to process images built
mainly using OpenCV. It could thought as the core of the project.
- RRtoolFC is a development tool using sequential function charts (FC stands for Function Chart).
This tool is still under development and can be thought as the programming interface.
- Imrestore is an application to restore images in general but in this case is configured to
restore retinal images. Because it is still under development it is currently used for testing and to let the users find out about its utilities.
Do not hesitate to share or use the project. Currently the project is under development and it is for research purposes.
Stable:
- Documentation: http://pythonhosted.org/rrtools
- Download Page: https://pypi.python.org/pypi/rrtoolbox
Latest:
- Documentation: http://rrtools.readthedocs.io/
- Project Homepage: https://github.com/davtoh/RRtools
BSD license, David Toro <[email protected]>
For API documentation, usage and examples see files in the "documentation" directory. The ".rst" files can be read in any text editor or be converted to HTML or PDF using Sphinx. A HTML version is available online at http://rrtools.readthedocs.io/en/latest/
Read RRtoolbox (Retinal Restauration Toolbox) manual or the dissertation which made this project possible with all the concepts and explanations.
Examples are found in the directory examples and unit tests in tests.
pip install rrtools should work for most users.
The usual setup.py for Python libraries is used for the source distribution. But in most cases OpenCV must be installed separately usually compiled from source. However RRtoolbox has a mock module for cv2 called cv2_mock to let the user use the functions that do not need OpenCV. Notice that this imports RRtoolbox.cv2_mock as cv2.
To install OpenCV on windows without much hassle I recommend installing the binaries from the Unofficial Windows Binaries for Python and for Debian distributions I provide the bash OpenCV linux installation so that the user can compile openCV (it can take some time). Bear in mind that for Linux it downloads the latest 2.9 version instead of the new OpenCV version 3 because it does not consent to using non-free sources. You must accept the terms for OpenCV 2.9.
Once rrtools is successfully installed you can import the toolbox in python as:
>>>> import RRtoolbox as rr
All releases follow semantic rules proposed in https://www.python.org/dev/peps/pep-0440/ and http://semver.org/
To create an executable from source code:
$ pyinstaller -p ./ -n imrestore -F ./imrestore.py/ --version-file=version
To test imrestore script:
$ python imrestore.py tests/im1*
To test imrestore executable:
$./dist/imrestore tests/im1*
To test imrestore executable if in dist folder:
$ ./imrestore ../tests/im1*
- A basic usage would be typing
./imrestore tests/im1* --lensin the terminal which tells - imrestore to load images from the test path that start with im1 specified by the wildcard
'*' and the option --lens which adds, as its name implies, lens to the retinal area.
- This will load the following image which presents some flares and noise:
Retina photo 1
- The second image is from a different perspective of the same retinal area but has information that the other does not have:
Retina photo 2
- And Voilà they are merged into one. Notice how the flares tend to disappear and the lens were added too! Of course, because it is still on development the images are not that pretty...
Retina photo result
For help just type in ./imrestore --help, it could not be easier than that! A demo
is available using the jupyter notebook to generate the desired commands to learn or use
in the console while still running the program. Feel free to email me if you need help with
the code or would like to have additional features.
- Contributions and bug reports are appreciated.
- author: David Toro
- e-mail: [email protected]
- project: https://github.com/davtoh/RRtools