Colorimetric and Spectral Combined Metric for the Optimization of Multispectral Systems

We present here a new combined metric which takes into account spectral, colorimetric and integrated irradiance similarity between two spectra to be compared. This approach does not avoid ‘mononumerosis’, as would be desirable, but does seek to find a balance between three different, and important points of view in spectral match quality. This permits us to work with a single but meaningful cost function in minimization problems using, for instance, simulated annealing algorithms. The annealing algorithm requires us to minimize a single cost function and thus we propose a solution that calculates both the spectral and colorimetric similarity of any pair of skylight spectra. We also compare it with some other widely used metrics in an experiment to recover skylight spectra.