Contiguity analysis is concerned with the ways in which pixels of a class are grouped together. Groups of contiguous pixels in the same class, called raster regions, or clumps, can be identified by their sizes and manipulated. One example of contiguity analysis is for locating helicopter landing zones that require at least 250 contiguous pixels at 10-meter resolution.
Contiguity analysis can be used to: 1) divide a large class into separate raster regions, or 2) eliminate raster regions that are too small to be considered for an application.
In cases where very small clumps are not useful, they can be filtered out according to their sizes. This is sometimes referred to as removing the salt and pepper effects, or Sieve. Clumps smaller than the minimum size are recoded to a value of zero (0). In the figure below, all of the small clumps in the original (clumped) layer are removed.
Eliminate function removes the small clumps by replacing the values of pixels in these clumps with the value of nearby larger clumps. The final clumps are then recoded using the Original Value attribute so that the output values of the remaining clumps are in the same range as the values in the original file which was the input to Clump.
Eliminate differs from Sieve in that Sieve simply recodes the small clumps to zero rather than filling in with neighboring values. Also, Sieve does not recode the clumps back to the original pre-Clump value.
Use Clump, Sieve, and Eliminate (GIS Analysis) functions in ERDAS IMAGINE Workspace or Clump and Sieve in Spatial Modeler to perform contiguity analysis.