Three or more materials are advantageously separated from dual energy data by using a material separation technique. To effectively separate material clusters, a density plot is introduced to automatically render cluster separations. Initially, the projection data optionally undergo data-domain dual energy decomposition. Then, the image data is plotted in a vector plot whose axes are the low HU values and the high HU values. For a given data point in the vector plot, a number of data points is counted within in a region of interest surrounding the given data point to generate a density plot where each point now represents a density level surrounding the data point. Thus, clustering of a certain material is visualized by a predetermined color assignment scheme. Furthermore, special image processing methods such as Gaussian decomposition are used to improve the accuracy of material separation. In addition, the HSL color model may be used for better visualization and to bring a new dimension in material separation display.