Synthetic Aperture Radar (SAR) Image Processing Toolbox for Python
Before installation, please make sure you have the following:
- SciPy. Comes with many Python distributions such as Enthought Canopy, Python(x,y), and Anaconda. Development was done using the Anaconda distribution which can be downloaded for free from https://store.continuum.io/cshop/anaconda/.
- OpenCV. If using the omega-k algorithm, OpenCV is required. Instructions for installing OpenCV for Python can be found at https://opencv-python-tutroals.readthedocs.org/en/latest/py_tutorials/py_setup/py_table_of_contents_setup/py_table_of_contents_setup.html#py-table-of-content-setup.
- Spectral. Needed to interface with .envi files. Can be downloaded here: http://www.spectralpython.net/
To get started, run the demos in ./examples in a python or IPython console or your IDE of choice. For sim_demo, feel free to adjust the parameters located in ./examples/dictionaries.
Current capabilities include modeling the phase history for a collection of point targets as well as processing phase histories using the polar format, backprojection, and omega-k algorithms. Autofocusing can also be performed using the Phase Gradient Algorithm. The current version can interface with AFRL Gotcha and DIRSIG data as well as a data set provided by Sandia.
To install, first download and unzip the repository. Then from the command line, go to the unzipped directory and type "python setup.py install". To uninstall, simply remove the ritsar directory. This can be done by "rm -rf /(Python Directory)/Libs/site-packages/ritsar" for an anaconda distribution of python.
Data included with this toolset includes a small subset of the AFRL Gotch data provided by AFRL/SNA. The full data set can be downloaded separately from https://www.sdms.afrl.af.mil/index.php?collection=gotcha after user registration. Also included is a single dataset from Sandia National Labs.
If anyone is interested in collaborating, I can be reached at [email protected]. Ideas on how to incorporate a GUI would be greatly appreciated. I would also be interested in adding a fast-factorized backprojection algorithm at some point.