Consideration and Experiments on Object Spectral Reflectance for Color Sensor Evaluation/Calibration

The influence of existing spectral reflectances and sensor sensitivities on color reproduction quality is studied. We analyzed the distribution of 17 categories of object spectral reflectance, using the Standard Object Color Spectra Database for Color Reproduction Evaluation (SOCS). Then we evaluated the color reproduction quality of sensors for each category using SOCS and five-sample sensor sensitivity, allowing linear color correction as a post-processing to sensor outputs. We obtained the following results: a.) Spectral reflectance restoration quality is closely related to the subspace dimensions that the spectral reflectances span. b.) The rank of sensor-set color reproduction quality is mostly stable and not greatly affected by the object categories. c.) If we choose a category of artificial object colors for linear correction parameter calibration; the parameters may be applied to other categories of such object. These findings suggest that the color reproduction quality of sensors can be estimated using some standardized spectral reflectance data sets.