EM 1110-2-2907
1 October 2003
(2) Image Mosaic and Image Subset. The mosaic process is a common
feature in image processing programs. It is best to perform image enhancements
prior to piecing separate scenes together. Once the images are pieced together, the
resulting image may be large and include areas outside the study region. It is good
practice to take a subset of this larger scene to reduce the size of the image file.
This will make subsequent image processing faster. To do this, use the clip or
subset function in a software program. The clip function will need to know the
corner coordinates of the subset (usually the upper left and lower right). Some
software may require this procedure to be repeated for each individual band of
data. The subset should be named and saved as a separate file or files. Note: An
image subset may also be required if the margins of a newly registered scene are
skewed, or if the study only requires a small portion of one scene. Reduction of
the spatial dimensions of a scene reduces the image file size, simplifies image
classification, and prepares the image for map production.
Example: Calculate the percentage of land cover types for a classification
performed on a Landsat TM image with a spatial resolution of 30 m using a
supervised maximum likelihood classification with a 3.0 standard deviation.
Solution: Calculate the percentage based on the total
Percent Calculation
Number
Class
Percentage
of class
pixels
Water
16,903
(16,903/413,469) 100 = 4.1%
Forest:
368,641
(368,641/413,469) 100 = 89.1%
Wetlands
6,736
(6,736/413,469) 100 = 1.6%
Agriculture
13,853
(13,853/413,469) 100 = 3.4%
Urban
6,255
(6,255/413,469) 100 = 1.5%
Unknown
1081
(1081/413,469) 100 = 0.3%
Total
413,469
(413,469/413,469) 100 = 100%
Maximum likelihood is a superior classifier and training classes are well
defined. This is evident in the low number of pixels in the unknown class.
Area can be calculated using the number of pixels in a class and multiplying it
by the ground dimensions of the pixel. For example the number of square
meters and hectares in the wetland class of this example is:
6,736 (30m)2 = 6.1 106 m2
Wetlands
606.24 ha
This last step is often not necessary as many software programs automatically
calculate the hectares for each class.
5-37