Details:
Summary:
A number of companies are offering CIELAB or CIELAB-like measurements with their RGB camera web viewing/inspection systems. Unfortunately, the capabilities and limitations of these systems are not always clearly identified and understood. This paper is an attempt to set reasonable expectations for CIELAB measurements derived from an RGB camera. A consistent terminology can help customers compare systems, so the standards are first reviewed for definitions of key words, such as accuracy, repeatability, and inter-instrument agreement. In the following section, overview is provided of what a camera-based color measurement system must do to in order to provide accurate color measurements. From these, it should be obvious that a simple profile cannot be expected to provide accurate color measurements from an RGB camera. A review of technical papers on the subject of the accuracy of RGB to CIELAB conversion is provided. In general, these papers concentrate on a single limiting factor, the spectral response of the camera. Even with the assumption that all other parts of the system contribute insignificantly to the error, the accuracy reported is still considerably worse than one would like from a color measurement device. How about the accuracy of reporting changes in color? Experiments are performed to determine whether an RGB camera can be used to accurately measure changes in color ( L*, a*, b*, and E). If these measurements can be trusted, then it would be possible to use an RGB camera to monitor color during a press run. CIELAB-like measurements could be made with the camera at color ok, and deviations from these initial values could be used for process control.