Glossary for Geospatial Science

boxcar classification or boxcar interpretation:   The simplest form of automated image interpretation whereby three data ranges are selected for three coregistered images (like red, green, and blue).  The three data ranges define a boxcar shape if plotted on three dimensional perpendicular axes that represent possible data values in red, green, and blue.  The ranges are usually selected to represent the color variation in the three rasters for a feature of interest (like all the dark brown areas representing bare soil).

All of the data triplets in the three raster objects are tested to determine if they represent a cell inside the boxcar.  All the inside cells may then be displayed or otherwise recorded as “classified.”

Boxcar classification is not restricted to sets of three rasters.  It may be used with any number: using 2 rasters designs a 2-dimensional “boxcar” test in a 2-dimensional space, and using “n” rasters defines an n-dimensional “boxcar” test in n-dimensional space.