How to Measure Paper's Reflectivity

Sciurus

Hi, everyone.

I would to identify colors (fur colors, in this case, but I suppose that's not important) with greater precision that using color words, (eg. Rufus). Ideally, I'd use a spectrometer to find reflectance curves, but I don't have one, so I thought that using a digital camera would be the next best option.

By comparing the RGB color values from a photograph to those of a piece of white paper, I can report the relationship of the three color channels to white paper, but is there a way I can determine the absolute reflectivity of the paper so I can report absolute values for the color channels?

Thanks for the help.

Andy Resnick

Paper (non-gloss) is a perfectly fine "white" standard for your application.

If you want absolute measurements, you need to standardize your illumination conditions (using a ultra-stable lamp) and optical system as well- paper is a diffuse reflector, so you have to control for both the solid angle of illumination and collection. Plus use something other than paper- Spectralon, for example. And then characterize the full BRDF curve, since fur is not a Lambertian reflector. And then, digital cameras have Bayer filters, so neighboring pixels are not directly comparable.

Basically, you are taking a perfectly valid, simple, and *easy* measurement technique (relative measurements) and converting it into a horribly complicated absolute measurement technique. And for what? an extra few percent precision?

Sciurus

Andy Resnick,

Thanks for the reply. I suppose fur has a larger surface area per area of paper, and that would introduce uncertainty. And since the topography of fur is different than paper, lighting conditions and photography angle would have to be standard. I gather that's what "not a Lambertian reflector" implies. Measuring color seems like is should be simple, but I guess it isn't.

Of course, I'm not looking for high precision -- fur isn't precisely homogeneous, anyway, so if there's no simple way to produce absolute values, I'll just keep working with "relative to paper" values, as you suggested.

But, just out of curiosity, is there a common household substance that has standard reflectance values?

Andy Resnick

A Lamertian reflector has no gloss reflection- that is, all the light is scattered uniformly in direction. The moon is (nearly) Lambertian, which is why it looks flat rather than spherical. Fur has a glossy component to the reflectivity.

I can't think of a household item with "standard" reflectances-I mean, a mirror is nearly 100% reflective, sugar and flour nearly 100% scattering, but in terms of a reflectance *standard*, I think there's too much variability in household objects. Seriously, basic white bond paper is an excellent reference.

Sciurus

Thank you, Andy Resnick. That makes sense.

klimatos

Using "color words" is not that bad a system for something as subjective as the color of fur. Assuming, of course, that you have an extensive color vocabulary.

For something a bit more objective, the old (1975) Manual of Remote Sensing, came with a beautiful set of color chips, a couple of hundred of them as I recall, The work was published by the American Society of Photogrammetry in Falls Church, Virginia. My copy has since gone astray, so I can't be more specific.

Another group that emphasizes gradation of color and tint includes stamp collectors. As I recall from my philatelic days, there are several hundred recognized philatelic colors.

Andy Resnick

That's a good point- numbers (even RBG triplets) aren't necessarily the best way to label colors. In addition to the schemes klimatos mentioned, there's also the Pantone system

http://www.pantone.com/pages/pantone/index.aspx

Munsell system:

http://www.xrite.com/top_munsell.asp...FQTNKgod5V9Q_Q

And more:

http://en.wikipedia.org/wiki/List_of...and_their_uses

Sciurus

That's true. Currently, when someone describes fur, as say, "white", there's no precision explicitly given. Now, there's a list of colors that are commonly used when describing fur, and you could assume that the fur was closer to white than one of the other colors, but that's an assumption, and I want to minimize the assumptions that my readers have to make. I'm not writing anything important, but it's more the principle. If there were a defined list of colors from which to choose the closest match, then I would at least be saying something with precision.

Probably color chips would be better, since using a camera adds more factors to the measured data. As Andy Resnick mentioned, the camera uses a Bayer filter. Are the filter characteristics the same from one camera to another? Shine will be a bigger issue with the camera, although I hope that the effect will average out if I take a large enough image sample. Also, the eye sees more colors than the camera can capture.

For now, I have a camera. And if I use a camera, I don't have to trust that my own eye is properly standard in its color perception.

Pantone looks like it might work, but it looks like I wouldn't be able to use it because its use is controlled. The chips are something I'll have to read into a little more. Thanks for the tip, klimatos.

Andy Resnick

The Bayer filter is an industry standard; all (consumer color imaging) sensor chips should be directly comparable. Shine (not to mention spots) will be a tricky problem to deal with. Shine can be controlled by suing a diffuse source, like a light box. Not sure how you can deal with non-uniform coloring.

I'm not sure that your eye can distinguish more colors than a camera can register, but that's an excellent point. Certainly, human vision covers more space than the RGB (television standard) can display:

http://hyperphysics.phy-astr.gsu.edu...ision/cie.html

but that's different than how much color space a camera covers. Human vision can distinguish up to 256 discrete levels of grey, this corresponds to 16,777,216 discrete (r,g,b) colors. However, camera sensors may have more than 8 bits of dynamic range which could be used to open up additional color space.

Sciurus

Non-uniform color will increase the standard deviation over my sample area, and I can report that as well, if necessary, I suppose.

Thank you for pointing out the error in my logic. I had thought that since a RGB monitor cannot display all the colors the eye can see, an RGB camera could not capture them. If the Bayer filters and photo doides had the same spectral sensitivity characteristics as the photosensors in the eye, a camera would. I don't think they quite do, though.

Here's a measured camera gamut. http://www.amateurphotographer.co.uk...fset=&offset=7

Thanks for the replies. Some of this may be outside of the scope of what I need to consider for measuring fur, but it's interesting to research, anyway. And perhaps in the future, I may actually need precision.

Andy Resnick

Thanks for that link- I was not familiar with the site. They exactly describe the white balance issues/artifacts my camera has, and also suggest ways to counteract it.

Claud Butler

Sciurus,

I found this thread by accident, and even though it is a bit old, I wanted to know whether you had solved your fur colour measurement problem? This is partly because I am a former squirrel ecologist and partly because I still have an interest in fur colour and its measurement.

There is a standard reflectace material used by photographers, simply called a grey card. I think Kodak are the only manufacturer. It has a nominal 18% reflectance, but you'd need to check their technical data sheet to know across what range of wavelengths this applies.