In this article, we present a novel treemap coloring method which can help users to analyze visual data more easily. Our method overcomes two major limitations of existing treemaps in that they are either aesthetically unpleasing or unable to readily discriminate data blocks with close sizes. Our study indicates that the use of proper color schemes can surprisingly address these two seemingly uncorrelated limitations simultaneously. To improve the aesthetic value of a treemap, we apply the color aesthetic model to treemap generation. To better the degree of data discrimination of similar data, based on the principle of expansive and contractive colors, we propose a novel quantitative color-visually perceived area (C-VPA) model via experimental methods. Furthermore, we combine these two models to derive a genetic algorithm-based treemap coloring method. Our experimental results confirm the superiority of our method in terms of improved data discrimination and aesthetics of the treemaps.