Satellite Contrast Methods

  Set satellite contrast.

Contrast enhancement allows you to emphasize different aspects of the image. With no enhancement the image will typically contain very little contrast, because few scenes use the full dynamic range of the sensor. The sensor can detect a specific number of radiation intensities (typically 256), and most scenes will not have the brightest (whitest snow) or darkest (pitch black basalt flow) scenery. The contrast enhancement algorithm takes the 256 brightness levels sensed by the satellite and transforms them to 256 colors used on the display screen, hopefully using all 256 (or at least the full range from 0 to 255).

This diagram shows an example of a histogram equalization. Because this scene has no pixels with very low or with high reflectance values, it would have no black or very dark grays, and no whites or very light grays. All pixels would be medium shades of gray. To use the full range of the gray scale available, the stretch assigns the DNs actually present in the image to the full range of DNs available for the output device. This usually results in all possible output DNs not being used, but the eye usually will not notice this nearly as much as it would if the dark and light shades were not used. If 256 gray shades are available but the eye can only differentiate about 30, the best results can be achieved in the colors actually used are spread as far as possible across the available range.

There are options to modify the satellite histogram on the Modify menu of the map window.

This table shows the same scene in central Missouri from an NGA CIB publicly posted on the WWW. It shows how changing the contrast method can improve or harm the ability to interpret the scene. No single method of contrast enhancement works in all cases, and you must experiment to determine what works best in a given situation.

No contrast enhancement--uses values in the file without modification. With CIB this results in no bright whites because of the unique characteristics of the CIB color palette which only uses 216 colors in a look up table.
Histogram equalization. Attempts to have the same number of pixels in each shade of gray.
Linear stretch--black is assigned to the lowest reflectance in the image, and white to the highest. Values in between are interpolated linearly. If these values at either end are not typical, very few pixels may appear in either the white or black end and most pixels will be middle shades of gray.
Linear stretch, ignoring 1% of the pixels at each end of the distribution. These 1% are shown in white or black. Clouds are one reason to use this stretch, and the percentage to ignore depends on the number of clouds (which ideally should be close to zero, but in certain mission critical cases you might have to use whatever imagery could be acquired).

Increasing what is ignored at the high (cloud) end lightens the image.  When clouds are present, this will have the largest impact.

Increasing what is ignred at the low end darkens the image.
Linear stretch, ignoring 5% of the pixels at each end of the distribution. These 5% are shown in white or black, which is apparent when comparing this image with the 1% tails.  In this case the 5% is probably excessive, and a lot of the developed area in the upper right is saturated white with no detail.

Last revision 12/13/2017