Hello,
I was wondering if you would have a suggestion of how to normalize the reflectance measures you obtain from the cone catch image by area (number of pixels), before even doing the QPCA analysis.
I.e., I have generated multispectral images from a few individuals combining photographs in the visible and UV spectrum. Then, I generated a cone catch image, where I delineated a ROI for each animal which is the dorsal region. At this point, I measured the reflectance mean and SD for UV, lw, mw and sw lengths. Although the individuals are from the same species, they have different sizes. I have mostly selected the dorsal region, but of course ROIs between individuals end up having different number of pixels.
Would be best to, for example, select a given amount of pixels for each animal, to make sure I am measuring roughly the same area? Or is there a way to extract the size of each ROI and then use that normalize the reflectance measurements? For this last option, do you have a suggestion on how to normalize (divide by area?). Also, any suggestion on how to extract ROI area using micaToolbox?
Thank you
Mafalda
Hi Mafalda,
The reflectance value of a surface does not change with its size, so normalising by surface area for reflectance values and cone catch wouldn’t make sense. However, calculating means, standard deviations etc. IS meaningful if there is variation in reflectance across a surface.
Please also note that, to generate weighted spatiochromatic representations of surfaces with variable colour and luminance contrast, the RNL clustering & VCA/CAA/BSA pipeline is tailored to correct for variability in size.
Cheers,
Cedric