Spatial analysis of pixelated data

Cluster ver. 1.0 performs a new type of spatial analysis of pixelated data (SAPD). Users manually select up to 24 non-target colors in a digital image and then move a slider bar to highlight the clusters of target pixels in the image. The app calculates the percent area of the image occupied by the identified clusters.

Before performing spatial analysis, the user deselects anomalous clusters by visual inspection and by setting the minimum threshold value for cluster size. A test for random distribution of clusters is performed by a t-test that compares mean inter-cluster distance in the observed image with a user-selected number of randomly generated images, wherein clusters are placed in rank order of size and both the exact shape and compass orientation of each cluster is retained.

Program output is a downloadable spreadsheet that includes data about the clusters, including cluster size, number, proximity, and cluster shape parameters.

The app may be applied to any digital image at any scale from the molecular to the galactic. Patterns analyzed can include those in ecology, geography, botany, demographics, astronomy and other disciplines.

Cluster was developed by the University of Hawaii at Manoa and Cornell University.

Release date: 15 November 2016