Public Lab Research note


GSoC 2014 - Automatic Spectrometer Calibration

by Sreyanth | May 20, 2014 17:29 20 May 17:29 | #10491 | #10491

Introduction

My GSoC project this year deals with enhancing the awesome SpectralWorkbench support automatic spectrometer calibration, improving the macros module and developing a self-contained offline SWB.

A detailed proposal of my project is found here. The initial discussions we did can be seen here.

My local repo of SWB: https://github.com/Sreyanth/spectral-workbench

An up-to date SWB can be cloned from: https://github.com/publiclab/spectral-workbench and from: https://github.com/jywarren/spectral-workbench

What happened till date?

I have setup the Desktop Spectrometry Kit (more details about the kit here) and tried capturing the spectrums. My trials can be viewed on my profile. I have compared my captures with the existing CFL spectrums and am quite convinced that my project is going to be of great help to the community (given that the quality of my spectrums is poor and difficult to calibrate manually).

I have also forked the latest version of SWB and am working on modularizing the SWB.

What happens next?

This note is only concerned about the automatic spectrometer calibration. There would be other 'progress notes' about macros module development and the offline SWB as well.

I am modularizing the SWB (most of the codebase is already modularized, but some extra modularization is required for smoother transition into the offline SWB development). Also, the common code for both the online and offline versions can be dealt with ease.

After this, I would implement a brute force method to auto-locate the spectral band and start my work on automatic calibration (including auto-recognition of CFL). I assume that Pascal's work will concentrate on auto-locating the spectral band, extracting the full-band RGB, correcting for curvature, HDR to extract data from clipped regions and produce a quality spectral response. (view his proposal here)

Have suggestions?

Any suggestions are always welcome. Either contact me directly, or my mentors Jeff and Dave. Alternatively, you can mail to plots-gsoc@googlegroups.com with the subject: "GSoC 2014 - Automatic Spectrometer Calibration - Suggestions". Also, you can simply comment on this note. :-)


2 Comments

The existing peak detection code, and notes on its shortcomings, is here: https://github.com/jywarren/spectral-workbench/blob/master/webserver/public/javascripts/api.js#L68-L181

One of the big problems was that if one peak were missed, the ratios wouldnt be recognized because they're order-dependent!

Reply to this comment...


I may be really off base here so play nice if I am.

What sort of literature search has occurred regarding peak detection? I did a search and immediately found this. I just suspect that there are a lot of other resources available that are useable. This area of research is broadly applicable and thus there has been a lot of work done on it.

Is this a question? Click here to post it to the Questions page.

Reply to this comment...


Login to comment.