Feature Clustering to Measure Clutter in Electronic Displays

Abstract : A technique is presented to cluster geospatial features on an electronic display and determine a meaningful measure of display clutter. An algorithm previously developed by the Naval Research Laboratory (NRL) to cluster objects in sidescan imagery has been modified to cluster any displayed features in three dimensions: geospatial location (x, y) and color (z). This paper presents preliminary results of the clustering algorithm and cluster density calculations for a series of electronic displays with varying levels of clutter. The clutter metric correlates with preliminary, subjective clutter rankings. Our next step in validating this method will be correlating the metric with user performance.