A Segmentation of the Landsat ETM+ Image displayed on the left is shown on the right. The new approach eliminates processing artifacts.

A further modification has been made in the algorithm and implementing software reported in “Modified Recursive Hierarchical Segmentation of Data” (GSC-14681-1), NASA Tech Briefs, Vol. 30, No. 6 (June 2006), page 51. That software performs recursive hierarchical segmentation of data having spatial characteristics (e.g., spectralimage data). The output of a prior version of the software contained artifacts, including spurious segmentation-image regions bounded by processing-window edges.The modification for suppressing the artifacts, mentioned in the cited article, was addition of a subroutine that analyzes data in the vicinities of seams to find pairs of regions that tend to lie adjacent to each other on opposite sides of the seams. Within each such pair, pixels in one region that are more similar to pixels in the other region are reassigned to the other region. The present modification provides for a parameter ranging from 0 to 1 for controlling the relative priority of merges between spatially adjacent and spatially non-adjacent regions. At 1, spatially-adjacent-/ spatially-non-adjacent- region merges have equal priority. At 0, only spatially-adjacent-region merges (no spectral clustering) are allowed. Between 0 and 1, spatially-adjacent- region merges have priority over spatially-non-adjacent ones.

This program was written by James C. Tilton of Goddard Space Flight Center. For further information, access the Technical Support Package (TSP) free on-line at www.techbriefs.com/tsp under the Software category. GSC-14994-1