Basically they are supposed to be an uncertainty metric (standard deviation) for the RGB measurements. But there are a few crucial points. IIRC, scanin fills the STDEV_R/G/B colums with the standard deviation of the RGB numbers of the pixes sampled from a patch. So basically they represent the amount of random noise present in the captured image at the location of the patch. This metric does however not directly reflect the uncertainty of the RGB_R/G/B numbers in the .ti3 file, which would be actually of interest. The latter basically consists of two components, (1) a random component which is however by a factor of sqrt(number_of_pixels_sampled_from_patch) lower than the sample standard deviation in the STDEV_R/G/B columns, and (2) unknown systematic errors (e.g. glare in case of a scanner/camera device) which are not covered here at all, though they can become quite dominant in practice. So I'd still treat the STDEV_... numbers with caution... Best Regards, Gerhard Am 27.09.2011 22:48, schrieb Matthieu DUBAIL:
I could only make a guess. I think the stdev might be a latitude from the intended target (deviation?). Within the latitude near the target color, the reading is considered as successful. It's useful for Chartread. Le 26 sept. 2011 à 09:38, tb<baruchel@xxxxxxx> a écrit :On Sun, 25 Sep 2011, tb wrote:Could you show to me such a "basic", clean and simple TI3 file (or cgats data txt file) for such a six-column-based set of data?OK, I finally could find some of them: http://www.freelabs.com/~whitis/reviews/230X_USB_Microscope/Camera.ti3 http://lj.vladimirovich.net/lj11/spectral/4500.ti3 But I have some more questions concerning the format: SAMPLE_ID XYZ_X XYZ_Y XYZ_Z RGB_R RGB_G RGB_B STDEV_R STDEV_G STDEV_B Could you confirm that columns XYZ_X, XYZ_Y, XYZ_Z are the "target" (the intended colors?) Could you confirm that columns RGB_R, RGB_G, RGB_B are the "seen" colors? I can't really figure out what are the three last columns: STDEV_R, STDEV_G, STDEV_B What do they mean? By the way, why can I find on several ti3 files on the web: KEYWORD STDEV_R KEYWORD STDEV_G without a line for STDEV_B ? Best regards, tb.
