Image Engineering camSPECS User Manual page 22

this database must be replaced with a database of reference spectral sensitivities from cameras of the
same make and model sampled from production before parameter optimization.
Description of the numbered items marked in the image above:
1. Spectral Sensitivity Database selects the reference spectral sensitivities that are representative of the
production cameras whose individual spectral sensitivities are to be determined. It is assumed that a
minimum representative sampling of 30 cameras has been measured previously with camSPECS as
described in section 4.5 item 8 or another monochromator and that these reference spectral sensitivities
have been saved in the correct SS_RGB.txt format. With this new representative spectral sensitivity
database selected, the parameters described next can be optimized.
•
To create the database, Folder selects all the reference spectral sensitivity data within the selected
folder.
•
File selects the reference spectral sensitivity database directly and is the recommended selection.
2. The Method is a pop-up that offers choices for how the spectral sensitivities of the test camera are
computed.
•
The MSE (Mean Square Error) is the simplest method and applies in most cases where the variance
in the spectral sensitivities database is small and arbitrary constraints are not required.
Regularization controls the tradeoff between the accuracy and robustness of the results. Use the
smallest value that minimizes the difference between the computed and the Reference Spectral
Sensitivities for the set of test cameras.
•
The PCA (Principal Component Analysis) method has increased precision when the variance in the
Image Engineering
Page 22 of 36