Weighted distance mapping (WDM)
Egon L. van den Broek · 2005
A new distance mapping technique is introduced: weighted distance mapping (WDM). It is based on an adapted version of Fast Exact Euclidean Distance (FEED) transforms. It computes, after assigning a metric, a probability space for partly categorized or clustered data. This is visualized by gradual intensity changes as illustrated by the categorization of a color space based on clustered data points. In addition, edge detection of boundaries between categories can be done to find exact borders of clusters or categories. Hence, Voronoi diagrams can be created. The proposed WDMs, with or without exact edges, provide a new rich source for data analysis as well as an intuitive method of describing structure in data.