Page 304 - Fundamentals of Light Microscopy and Electronic Imaging

ADJUSTING GAMMA (Γ) TO CREATE EXPONENTIAL LUTS 287

brighten or darken an image, the white and dark set handles are moved, respectively, to
the left or to the right.
Before making any adjustments, the user should first get to know the image by
determining the pixel values and locations on the image histogram corresponding to var-
ious regions in the object, including shadows (or dark background in the case of fluo-
rescence images), bright features (called highlights), and other features of intermediate
intensity. This is done by moving the cursor over different regions of the image and
reading the corresponding numeric values from a statistics or status window. Knowl-
edge of the intensities of different objects and their position in the histogram will allow
the user to act more judiciously and conservatively in selecting the black and white set
points to be used in histogram stretching. This is also a good time to check the image for
indications of clipping, the appearance of saturated white or underexposed black areas
in the image. Clipping should be avoided, both during acquisition and during process-
ing. With this knowledge in mind, the operator then sets the handles on the histogram.
An objective and conservative approach is to include all of the pixel values representing
the specimen within the upper and lower set points. It is useful to remember that human
vision cannot distinguish all of the 256 gray-scale values present in an 8 bit image, let
alone a deeper one. Therefore, as the situation demands, you are justified in altering the
display by histogram stretching to see important internal details. Guidelines on the use
of histogram stretching are presented in Chapter 16.
There are several ways of displaying the image histogram. The standard or regular
histogram has already been described and is shown in Figure 15-2a. A logarithmic his-
togram shows the input pixel value (x-axis) vs. the number of pixels having that value
on a log scale on the y-axis. This type of display is useful when it is necessary to see and
consider pixel values that comprise just a minority of the image and exhibit a strong
response to histogram stretching (Fig. 15-2b). This should not be confused with a loga-
rithmic LUT, which uses logarithmic scaling to display the image itself.
Another useful histogram display is the integrated or cumulative histogram, which
can be convenient for adjusting the contrast and brightness of certain images (Fig.
15-2c). This histogram display is more useful for phase contrast, DIC, and bright-field
images that tend to have light backgrounds than it is for dark fluorescence images,
which typically include a wide range of intensities. In this histogram, the x-axis shows
the input pixel values, while the y-axis shows the cumulative number of all pixels hav-
ing a value of x and lower on the x-axis. For relatively bright images (DIC and phase
contrast microscopy), the max-min handles (white set/black set handles) are usually
moved so that they define a line that is tangent to the rapidly rising slope of the his-
togram.

ADJUSTING GAMMA ( ) TO CREATE EXPONENTIAL LUTs

As mentioned, LUT display functions can be linear, logarithmic, exponential, or, in the
case of Adobe Photoshop software, even arbitrarily curvilinear as set by the operator.
Linear functions (in byte format) have the form

Displayed value 255 (Data value Min)/(Max Min),

and are commonly used for adjusting the image display. However, linear functions pre-
sent difficulties when the goal is to include all of the pixel values contained in the image.