I'm developing a software (.NET/Mono) library to work with color spaces and light spectrums, and I want to add many test cases in order to ensure software quality.

I've programmed code to convert light spectrum samples to the 1931 CIE XYZ color space (applying a convolution and all this stuff, all explained in the Wikipedia).

My problem is I can't find (light spectrum, XYZ) pairs in order to test my code, anyone knows if such data exists and where can I find it?

P.D.: If anyone is interested on the library, it's in the following URL: http://github.com/Litipk/ColorSharp

  • $\begingroup$ Strictly speaking, this is not a physics question. The response of the human eye to a spectrum is a problem of physiology, which is complicated by the difficulty to derive such a response function from a genetically inhomogeneous population (not all people have the same color vision, nor is color vision a guaranteed evolutionary constant). The CIE XYZ color space is a definition that can only be tested against an independent clean room implementation of the document, but not against physical data. What interests me is what the library is used for? Could you given an example? Thanks! $\endgroup$ – CuriousOne Oct 15 '14 at 19:18
  • $\begingroup$ Yes, it's true that this question is not about physics (but I couldn't find a better forum inside Stack Exchange and I've seen that there are many CIE XYZ questions in this forum, I'm sorry). The library will perform conversions between colour spaces, including from light spectrum to concrete colour space, and from colour to light spectrum (adding restrictions, like efficiency or "colour quality", because there is more information in the spectrum form than in the colour space form). It also will compute features like "colour temperature". $\endgroup$ – castarco Oct 15 '14 at 19:31
  • $\begingroup$ Thanks for the update. I can see why it would be useful to augment the usual technical color spaces with a proper physiological model. My feeling as an experimentalist is that there is very little physics can do to validate this kind of software. You could certainly run a "standard" solar spectrum trough the tool and see if that adds up to "white" (?) in the XYZ space (is that even how it works?). The problem is that even an observed solar spectrum probably has larger error bars than the numerical error bounds on your library should be. $\endgroup$ – CuriousOne Oct 15 '14 at 19:38
  • $\begingroup$ @CuriousOne The CIE 1931 XYZ color space has very concrete matching functions, and also the majority of other standard color spaces. Yes, color perception is subjective, but color spaces have very specific definitions (based on empirical data: aggregating perceptions of many people). And I know how to do the transformations, but because I have to test software quality, I need external results (what If I'm wrong and I'm computing the transformations in the wrong way? I need to check it in an automatized way, using data sets computed by other people)- $\endgroup$ – castarco Oct 16 '14 at 8:46
  • $\begingroup$ I can see your problem... you need test vectors. Ultimately they will have to come from a cleanroom implementation, I suppose. Does the standard provide any examples of the required output, like for black body radiation of different temperature or for a standardized solar spectrum? $\endgroup$ – CuriousOne Oct 16 '14 at 14:36

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.