The next thing we'll do is scale so that the maximum linear RGB value is 1.0. We want to modify those curves so that they fall within the range and produce a reasonably accurate, smoothly varying, and attractive representation of the spectral colors. Title( 'Linear RGB values for spectral wavelengths') RGB_lin = xyz2rgb(XYZ, 'ColorSpace', 'linear-rgb') The Image Processing Toolbox function xyz2rgb can optionally convert to linear values. The typical RGB values we see for image pixels are related nonlinearly to light intensity, and linear RGB values are more appropriate for the following averaging and scaling steps. Work with linear RGB valuesįirst, let's convert the XYZ values to 'linear' RGB values. The method is a variation of the one described in: Andrew Young (2012). I'm going to show you the method used by the DIPUM3E function spectrumColors. These out-of-range values are being clipped to the displayable range, and that leads to a bad results. Everything outside that range (negative values, or values greater than 1), can't be displayed exactly. The gray shaded region shows the range between 0 and 1 this is the displayable range of colors. (The plotrgb and shadeGamutRegion functions are down below.) close What's going on with this? We can see the problem if we draw a line plot of the RGB values. It seems uneven, with several patches that seem to be mostly one color. That doesn't look like a very good spectral color plot to me. It uses the DIPUM3E function colorSwatches.) Let's try simply converting these XYZ values directly to RGB using the Image Processing Toolbox function xyz2rgb. Title( 'XYZ values for spectral wavelengths') Or we can find the XYZ values for all of the wavelengths we are interested in. To find the XYZ values for a specific lambda, such as 500, we can use interp1: interp1(T.lambda,500) But it is really simple: just interpolate into the CIE XYZ color matching functions, which are returned by the DIPUM3E function colorMatchingFunctions. This computation can be found in the DIPUM3E function lambda2xyz. The first computational step is to find the XYZ values for each value of lambda. Billmeyer and Saltzman's Principals of Color Technology, 3rd ed., John Wiley & Sons, NJ. The limits used here, 380 nm to 780 nm, are those given in Berns, R. What should the x-axis limits be on this plot? In other words, what is the visible wavelength that we are interested in? You will see different limits being used in different places. Convert to nonlinear RGB values for the final result.

Find XYZ values for the spectral colors.

Matlab color2gray