Workflow — Algorithm development, validation and source control protocols

We use SVN for version control of ComparisonAnalysis and HyperspecLab — our in-house image analysis framework, and all members of the group should review the summary below, as well as read some basic documentation about SVN. Minshan has created a useful summary of SVN functions, including some preliminary basics of SVN  (and using Trunks and Branches). Please review the same at: Some Useful Excerpts from SVN tutorials

Our lab environment involves 3 broad types of software:

  • ComparisonAnalysis: This is our internal software toolkit to benchmark algorithms, validate classification results with prior established approaches and so on. Any new algorithm we develop, after sufficient polishing and cleaning up will find it’s way to ComparisonAnaysis. This toolkit (ComparisonAnalysis) is internal to our group, and is not public.
  • HyperspecLab: This is a graphical user interface for loading, sorting, labeling and analysis of hyperspectral data. After our algorithms have reached sufficient maturity, and they are no longer “work-in-progress”, they may be considered for addition to HyperspecLab. The executable version (not the source) of HyperspecLab will be distributed through our lab’s website for collaborators and other end-users who want to test and use our algorithms for their scientific use of hyperspectral datasets.
  • Ongoing work: This reflects new algorithms we are developing for our projects, and for our thesis and dissertation research. Eventually, when they are implemented, they must be added to ComparisonAnalysis.

Information to be added later — use of NetCDF as a data format for geosensing data (Note Matlab provides native support to netCDF via functions such as ncdisp, ncinfo, ncread and ncwrite. In the future, we will be incorporating netCDF support to both HyperspecLab and ComparisonAnalysis.

 

This entry was posted in Hyperspectral, Matlab, SVN. Bookmark the permalink.

Comments are closed.