Robust Processing of Color Target Measurements for Device Characterization

Device characterization typically involves generating one or more targets each comprising a number of color patches and making colorimetric measurements of these patches. In order to derive the characterization function, one has to obtain the correspondence between the set of colorimetric measurements and a set of device signals. In the case of output device characterization, the device signals are the CMY(K) values used to create the target; for input device characterization the device signals are the captured RGB values. Establishing the correspondence requires knowledge of the spatial target layout, the order in which the measurement instrument measures the patches, and the order in which the input or output device captures or renders respectively the target patches. Since the measurement process requires human intervention, errors can conceivably be introduced at this step. A common example is misorientation of the target on the measurement stage. This results in an incorrect correspondence between device and colorimetric signals, thus producing an erroneous characterization. This paper presents some practical approaches to detect and correct for measurement errors due to misorientation. One approach reserves a number of patches in the target to encode the target layout and orientation. The measurements from this so-called “patch code” are then decoded to retrieve, and if necessary, correct the orientation. Another approach anticipates common misorientations of the target, and uses a pattern matching approach to select the correct orientation. This is done either automatically with a prior device characterization, or with visual validation by the user. Both techniques are successful in correcting for target misorientation with minimal burden to the user.