Need ideas for creating library

[gatohaus]

Firstly, thanks for SPy!

I've been hunting around trying to find a good way to build a spectral library, and haven't had much luck.

• data should be generally accessible and easily transferable (not platform or program specific, certainly not buried in a db)
• adding metadata or contextual info should be easy (tags, pics, notes)

At the moment the only idea I have is to use hdf5 files from python.
Imagine a search going through the files pulling spectra into memory to form a working library for the moment's specific task. (initial search time is not important) The pulled spectra would then be used to search through hyperspectral imagery for matches.

Anyone have a better idea?

• size wise it's something that would grow over time, but still pretty small.. I'd guess an average of 50 new spectra per month is about what is collected.
• my background: imaging systems engineer, long-term idl/envi user, mid-level python user

[Thomas Boggs]

I don't think I have a good answer for you but I'm looking forward to hearing what you ultimately decide is the best solution.

Making the data easily accessible and having the ability to add things like tags, pictures, & notes makes me think of a content management system. The Specchio system comes to mind but I get the impression that you want something a bit more self-contained / stand-alone.

You mentioned HDF5, which is probably a good option to consider. But I wonder then what level of ability you would want to query the data / metadata. To that end, you might also want to consider PyTables, which is built on top of HDF5 but adds some additional capabilities (e.g., a more robust query capability and support for data types not in h5py).

Please post an update when you find the best solution!

[gatohaus]

I hadn't heard of Specchio before, very interesting. Probably a good solution further down the line. But yeah, for now I need minimal infrastructure.

If I figure something out, I'll be sure to post here. :-)