Public Lab Research note


CFL detection by contrast, for auto-calibration

by warren | October 22, 2014 21:56 | 71 views | 0 comments | #11296 | 71 views | 0 comments | #11296 22 Oct 21:56

What I want to do

I was thinking we might be able to do a crude but very fast CFL detection by measuring the amount of contrast in spectra, so I wrote a script (could have used the API but wanted it to be loaded by default) to try it out. I ended up being a little more complex once I tried it out on a lot of spectra; I discarded sections with low slope, too.

continuous.png

My attempt and results

So it kind of measures how spiky a spectrum is, and does an OK job! It could use some tweaking and testing. My thought was to run this "cheap" method, then run a more complex matching if we have higher confidence that it's a CFL.

Try it out at https://spectralworkbench.org/capture (you may be prompted to reload the page once the cache clears) -- there's a little "lightbulb" button that will turn red if it thinks you're looking at a CFL.

Screenshots would be nice either if it works well or if it doesn't, to debug.

Questions and next steps

@sreyanth - does your code have a "degree of confidence" that it's found a CFL? Like, say it finds blue and green, but red doesn't match up. Is there some way to have a 'threshold' where it's like "yes, this is a CFL"?

Why I'm interested

In any case the dream is that the capture interface auto-detects a CFL and prompts you to auto-calibrate (the latter using @sreyanth's new code)


0 Comments

Login to comment.

Public Lab is open for anyone and will always be free. By signing up you'll join a diverse group of community researchers and tap into a lot of grassroots expertise.

Sign up