Paper on Practical Spectral Photography

[Brad Evans]

I found this paper titled "Practical Spectral Photography". Aside of their design, it contributes some quality validation science and an alternative for comparison.

http://www.cg.tuwien.ac.at/research/publications/2012/Habel_2012_PSP/Habel_2012_PSP-Publication.pdf 

[Jeffrey Warren]

wow, all of section 5 has good ideas for spectral and exposure calibration!

We should ask for their raw data, and maybe we could send them a spectrometer and they could use it on their pre-calibrated halogen lamp (that's probably expensive? but i really don't know) -- see text below.

Brad - would you be able to put up a research note on this so we can follow up on these techniques once people start getting their spectrometers? Maybe tag it "calibration"?

Awesome.

Jeff

-------------------------------------

Text:

5.2. Spectral Response Calibration 

The purpose of the spectral response calibration is to deter- 

mine the spectral sensitivities of each color channel. The 

spectral response can be determined by photographing a 

light source that has a spectrum that is continuous and known 

with high accuracy. For this purpose, we use a dedicated 

calibration halogen lamp. Note that this is the only step in 

the prototype construction that requires the application of 

previously calibrated equipment. We average over a few re- 

sponses and then simply divide through the known spec- 

trum of the halogen lamp. By considering red, green and 

blue Bayer pixels separately, we arrive at the spectral re- 

sponse functions Sr,g,b (λ) of each channel. We assume that 

the high-frequency structures of the response functions are 

noise and filter slightly with a 99% Fourier low-pass filter. 

The results of this calibration step are displayed in Figure 7, 

showing the spectral response of the complete optical sys- 

tem. 

On Tue, Sep 11, 2012 at 4:44 AM, Brad Evans <br...@belgaroo.com> wrote:

I found this paper titled "Practical Spectral Photography". Aside of their design, it contributes some quality validation science and an alternative for comparison.

http://www.cg.tuwien.ac.at/research/publications/2012/Habel_2012_PSP/Habel_2012_PSP-Publication.pdf 

-- Post to this group at publicla...@googlegroups.com. To unsubscribe, email publiclaborato...@googlegroups.com. Options at https://groups.google.com/d/forum/publiclaboratory?hl=en

[Mathew Lippincott]

5500K would be sunlight's color temperature, but getting black body (halogen) lights at that color temp is hard.  most high color temp bulbs are either LEDs or more than 500W-- and we need a light, not a grill.

Eiko SoLux bulbs are 4700k, and the spec sheet provides spectra.  They're all 12v halogens, so they aren't too pricy to mount.  B&H has the 35 watt bulbs listed for $12.50.  I don't know how good their reference spectra will be, as halogens tend to change color slightly over their lifetime.

[Jeffrey Warren]

Could we get relative differences between cameras by pointing several at the same light source at the same time?

[Brad Evans]

Great suggestions Mathew and Jeffery... 

I will follow up with a research note to focus the discussion. Ideally, we could address the grating and slit choice (as it influences attenuation and spectral resolution) and alignment of the grating together with calibration of the sensor for setup and ongoing use for absorption and reflectance spectroscopy.

Could we get relative differences between cameras by pointing several at the same light source at the same time?

You probably could, however, alignment (of the grating to the CCD/ CMOS pixels) and calibration would be much easier one at a time... and with CMSO/ CCD based spectrometers used in a range of environments sensor sensitivity becomes a function of temperature, this affects the dark current noise and essential means, calibration is needed for each spectra. 

Any chance there is already a thread on this I have missed?... 

Cheers,

Brad