Revisiting Perceptually Optimized Color Mapping for High-Dimensional Data Analysis

Color is one of the most effective visual variabks since it can be combined with other mappings and encode ilifonnation without using any adtational space on the display. An important exampk when expnssing additional visutJJ dimensions is dinly neetkd is the analysis of high-dimensionJJl dtlta. The property of perceptual linearity is tksirabk in this applicatio1t, because the user intu:itively perceives clusters and nlationships among multidimensional data points. Many approaches ust~ two dimensional colo1711Dps in their analysis, which an typica/ly cnated by interpolating in RGB, HSV or CIELAB color spaces. These approaches s'lum! the problem thai the re.ndting colon are either saturaled aniJ discriminative but rwt perr:eptuallinear or vice versa. A solution that combines both advantages has been pnviously introduced by Kilski et al.; yet, this method is to dQJe undenmlized in Injormalion Visualization according to our lilerature an.alysis. The JMthod maps high-dimension.al dtlta points into the CIELAB color space by maintaining the rttlalive perceivt~d distances of data points and color discrimination. In thi.J paper; Wtl generalize aniJ extend thll method of KtJski et al. to pTDVide perceptulll unifonn color mapping for vislllll tuUJlysis of high dimensional data. Further; we evaluale the method and provide guidelines for diffirnmt cmalysis tasks.