Contents || Index
04/22/2005

Scaling out of range data

  1. Load an image
    1. Help / Samples & Info / Images / Rutile 16 bit
    2. This is a 16 bit image with a few pixels with values much greater than the others, which causes the linear scaling to display an image that is black except for a few white spots.
    3. (How to view this image in Photoshop).
  2. Rutile image as loaded.  
    Note the three white pixels just to the left of this text.

    Info / plot histogram / original.  (This takes a long time to plot).

    Plots / log of front This is a logrithm of the plot on the left.  (This also takes a long time to plot.) (Log of zero is plotted slightly below the log of the smallest positive value).  Note the little 'blip' on the right that corresponds to the out of range pixels.  The next brightest pixels have values less than 4500.

     

  3. Scale using trimmed scaling. (There are other scaling options).
Scale / trimmed / .01 Scale / trimmed / 1  (default)

Info / plot histogram /scaled

(The scaled histograms plot quickly - there are only 256 values to plot.)

Info / plot histogram /scaled

 

Histogram equalization is another way to display the image.  Scale / equalize.

Info / plot histogram / scaled  Note that the histogram should be flat, ie. there should be equal numbers of pixels of every intensity.  The combed nature of the histogram is due to the pixel values being integers, rather than being a continuous distribution of real numbers.