|
Crop
Detection
Can you identify one crop from another in mixed fields?
This is a scene of Texas farmland. There are approximately 1,400
active fields in this scene, but our client was only interested
in finding one specific crop. In this test image, there are only
three fields of interest. Can you distinguish them?
IMAGINE
Subpixel Classifier did!
Objective
A seed producer was looking for a method to more accurately assess
acreage of a specific crop under cultivation. Their interest is
in monitoring the cultivation of this crop in different regions
throughout the world.
Includes material © Space Imaging L.P.
Approach
IMAGINE
Subpixel Classifier was used to exploit the time-cost advantages
of satellite imagery to identify crop locations in mixed environments,
in many countries, using a pair of comparison signatures.
Project Challenges
· This crop is often planted in remote areas
interspersed over large tracts
· On-the-ground survey over large areas is nearly
impossible
· High-resolution airborne imagery which might be
able to discriminate the crop is prohibitively expensive
· The signature had to be able to process scenes in
many countries including Brazil and India where the crop is grown.
Solution
IMAGINE
Subpixel Classifier was used to accurately detect the three
fields in this scene of Texas farmland.
Includes material © Space Imaging L.P.
IMAGINE
Subpixel Classifier’s unique Environmental Correction feature
allows portability of signatures from one scene to another. The
signatures used to detect the crop in this scene have been successfully
applied to fields in Kansas, Mexico and Brazil.
Technical Notes
Discriminating this crop from other crops is very difficult. An
approach was developed that employed two signatures (leaf-oriented
and stem-oriented) and multi-date imagery to detect only these fields.
In this instance, the three fields were detected—no other fields
were mistakenly identified. Even had there been some false alarms,
it was determined that the IMAGINE
Subpixel Classifier was a valuable tool for use in surveying
crop detection on a global scale.
|
