Can you distinguish the trees from low shrubbery and grasses?
As part of a military exercise in an arid region of New Mexico, military analysts used IMAGINE Subpixel Classifier as one of their tools for Intelligence Preparation of the Battlefield. Their goal was to identify areas where large military vehicles could be hidden...and could not be hidden. (See Tree Cover Mapping)

IMAGINE Subpixel Classifier did!

Objective
Analysts wanted not only to distinguish tree cover from low shrubbery and grasses, but also to determine tree cover density. Their objective was to maximize the effect of limited monitoring resources over a wide geographic area.

Challenge
Traditional classifiers and techniques can detect vegetative regions but can neither distinguish between various classes of vegetation nor describe the amount of vegetation within a pixel.

Maximum Likelihood Classifier
Note confusion between
vegetation types.
Also, no information regarding
amount per pixel.
Includes material © Space Imaging L.P.

IMAGINE Subpixel Classifier
Distinguishes trees from
other vegetation and provides quantification per pixel.

Includes material © Space Imaging L.P.

IMAGINE Subpixel Classifier
Tree classification of Pedernal Peak
Includes material © Space Imaging L.P.

IMAGINE Subpixel Classifier
Each shade represents a different
amount of tree cover per pixel

All green areas (IMAGINE Subpixel Classifier detections) that are visible under the yellow areas (Maximum Likelihood Classifier results) are tree covered areas with potential hide sites that were not detected with the traditional image classifier.

Another view of the difference between Maximum Likelihood Classifier results in yellow on the left below and IMAGINE Subpixel Classifier detections on the right in green below. The different shades of green represent the different density of tree cover reported in each pixel by IMAGINE Subpixel Classifier.

Maximum Likelihood Classification

IMAGINE Subpixel Classifier Classification