Using shape to visualize multivariate data

Abstract

This paper describes our recent findings in the area of using glyph shape to display one or two data dimensions in the visualization of 3D scalar and vector fields In our glyph-based visualization system, each glyph represents a data point in 3D space Visual attributes such as size, orientation, color and transparency can be mapped to data dimensions in the 3D space We are exploring the use of glyph shape as a display dimension, using superquadric superellipses as a means of supplying a parameterizable shape A basic factor in effectively using shape for quantitative visualization is determining how many (and which) superellipse shapes people can distinguish Since the superquadric shape's parameter set is not perceptually linear, we conducted a user study to which shapes people can generally distinguish The findings show that with large superellipses, about 22 separate shapes can be distinguished on average These results provide the foundation for exploring how effective superellipses may be in quantitative shape visualization