Kinesense: The Clarification Chronicles continues with Histograms this week. A histogram is a graphical representation of the tonal distribution in an image. In Kinesense there are three different options that can be accessed from the histogram clarification
Histogram Edit provides the selected transform algorithm with the same target values for each colour channel. You have the option to edit these channels individually. Moving the sliders on the bottom to just include the concentrated information will clear up the image. You also have the option to use either Linear or Histogram transform.
This option can bring out general detail in an image. It is a pre-configured algorithm that will attempt to stretch the contrast of the image via an analysis of it’s histogram. It differs from contrast stretch by using a non-linear transform function.
Histogram Local Equalisation
Local Histogram runs a small window over the image and carries out the equalisation process on each section of the image – this means that if you have an image with both light and dark regions, each region will be equalised separately. This is used when you wish to examine areas of over or under exposure within a scene for additional detail where a general histogram equalisation is insufficient.
Radius = size of the window
Pyramid Weighting = This defines the profile of the window. The standard option is to treat all samples within the box equally when calculating the histogram. Pyramid weighting gives more importance to the samples closest to the pixel. At a Pyramid Weighting value of 1 the sample co-located with the pixel has a value of 2xRadius, and a sample at one Radius distance has a value of one Radius. Lowering the Pyramid Weighting increases the importance of the samples closer to the pixel.
Pyramid weighting provides some smoothing to the HLW effect and can reduce the effect of haloing.
Global Peak = Limits the stretch to an image wide max. Good when the image has both bright and dark regions.
Diamond Sample = Window is rotated 45 degrees
Filter Histograms = The histogram modification uses the peak value found within the sample to set an upper limit. Filter Histogram removes the top n percent of samples from this calculation. The purpose of this is to mitigate the effects of noisy pixels on the calculation and is particularly effective when there are noisy pixels in an otherwise homogeneous area. NOTE: Global peak and Filter Histogram are mutually exclusive. Filter Histogram is performed before Contrast Limiting.
Contrast Limiting = Contrast limiting smooths the histogram by trimming any scores above a given value and redistributing the excess across the rest of the histogram. The intended effect is to reduce the effect of peaks from areas with a strong homogeneity.
For more in depth reading on this subject check out Histogram Equalization
Next week we’ll have a quiz to see if you’ve been paying attention!