Flatten
The Flatten module attempts to deal with uneven lighting within an image. Images are taken in low light
situations such as with flash or with cameras with uneven lighting distributions or scenes with an overhead
cast shadow are uneven in lighting distribution throughout the image. This significant artifact revels itself
once you attempt to segment the image using the flood fill, threshold, RGB Filter, etc. modules in that the
segmentation appears uneven across the entire image as the lighting changes.
The traditional technique used to deal with such situations require averaging over a large area of the image
followed by subtraction of that average from the original image. This technique was illustrated in our
Surveyor Trail following tutorial. As this technique
works well for many images that have highlights or large shadow areas this technique with several refinements
has been added as a distinct module.
This module is particularly useful when removing highlights or shadows. The examples below show flash light
correction, shadow removal, highlight correction and another shadow removal.
Interface
Instructions
1. Sample Area - The sample area defines how much the a localized area the module should look at when determining
the appropriate region intensity. Decreasing the size will reduce the image to the result of an edge detection
method, whilst increasing the sample area will focus more on larger areas of the image.
2. Automatic Intensity - The automatic intensity mode will attempt to produce an image that averages in intensity
at 128. This creates a intensity image that allows for good contrast.
3. Ideal Intensity - The Ideal Intensity mode will check the image for the most common intensity mode and produce
a final image that is close to that intensity mode. This is useful if your image appears too dark using the
automatic mode. Note that this mode can become confused and produce an overly dark or bright image.
4. Manual Intensity - There are times when both previous intensity modes will fail. At that point you can select
the manual intensity mode to produce images that appear correct.
Example
See Also
Equalize
Normalize
Color Balance
|
