Research On Digital Device Color Gamut Boundary Description And Evaluation Algorithm

Device gamut is a set of color which a device can render, the description of device gamut is very important in color gamut mapping and Hi-Fi color reproduction. In order to get an more accurate gamut, several Gamut Boundary Description (GBD) algorithms have been presented, and the researches is developing all the time, however there still many problems exist in the current GBD algorithms e.g. the accuracy, the consumed time, it is difficult to get an accurate gamut. Since there is not exist an universal accepted GBD evaluation algorithm, it is difficult to determine which GBD should be used in the color management applications.In this paper we present a new GBD algorithm using uneven segmentation method based on the research and analysis of the current algorithms. In the uneven segmentation method the device color gamut is divided into the saturate color parts and the unsaturated color parts, the saturate parts are divided into more segments through the CIEa*b* plane when calculating color gamut descriptors. After that, in each segment, the radius between the color points and the center point are calculated, for the color points which located in the saturated parts the color point with the biggest radius is selected as the gamut boundary descriptor, for that in the unsaturated parts, if the corresponding outer segments are empty the color point with the biggest radius is selected as the gamut boundary descriptor, if the corresponding outer segments are not empty, use the corresponding saturate gamut boundary descriptors as the gamut boundary descriptors. Then the gamut boundary descriptors are visualized in CIEL*a*b* color space using a triangulation method. Finally the uneven segmentation method is evaluated in contrast with other GBD methods using gamut mismatch indices (which can also be used to evaluate gamut similarity among different gamuts). Through a series of evaluation experiments (we have evaluated 68 device gamuts of 17 devices), results show that the uneven segmentation method performs better in accuracy and stability comparing with other current GBD algorithms.