Another question from me!
For a handful of the species for which we collected multispectral images, we also have point spec measurements of specific color patches. We’d like to match up normalized reflectance measurements of those same patches with the spectra to show that the multispectral images are capturing the relevant properties of each color patch (to ward off reviewers asking why we didn’t spec every individual on top of photographing it). Is there a simple way to do this? I was hoping to be able to do it from the polynomial models generated by the Generate_Cone_Mapping_Model_From_Chart.ijm script, but I’m not totally confident that makes sense in this case. Thanks in advance for any help!
The only way to compare the processes easily is post-conversion to cone-catch or hue data.
i.e. measure cone-catch with the camera and spec separately, and see whether they match well.
However, it’s not quite that simple. Spectrometers and cameras typically work differently in terms of light source. If your spec used a light source attached to the probe, then it’s measurements will be quite different in overall luminance (the closer the probe, the higher the reflectance), when compared to the camera. So the overall reflectance of the two measures might vary (substantially). So it might be better to use a hue metric of some kind (i.e. colour data that eliminate overall brightness). Converting to RNL chromaticity XYZ coordinates will be a good way to do that.
A further thing to check for is fluorescence. Pretty much everything is fluorescent to some degree, and the difference in light emissions spectra between your camera and spec-based measurements can create measurable differences (sometimes quite large differences).
As such, when you want really accurate colour reproduction between animal, camera and spectrometer I always recommend using photos and specs that are taken under natural sunlight (with a collimating lens on the spectrometer so that it acts just like a camera in terms of radiometric measurement). This is also made much easier if the things you’re measuring are are flat and diffuse (hence why I recommend colour charts or pastel sets).
Hope that helps,
Jolyon