Page 306 - Fundamentals of Light Microscopy and Electronic Imaging

FLAT-FIELD CORRECTION 289

response of the eye or camera film and are created using a function with variable expo-
nent such that

Displayed value [(Normalized value Min)/(Max Min)] .

The adjustment of and a figure showing LUTs with various values were pre-
sented in Figure 13-7. With settings 1, low pixel values are boosted relative to high
values and appear a medium gray in the image; this adjustment also reduces the contrast
between bright features and the darker background. A setting of 0.7 approximates the
response of the eye, allowing the image to more closely resemble the view we perceive
when looking in the microscope. Conversely, values 1 depress dark and medium
gray pixel values and increase the visibility and contrast of bright features. The effect of
adjusting on the image display is shown in Figure 15-3. In Photoshop, the slider used
to define the midpoint of gray values performs an identical function. Photoshop gives
you the choice of reading off a value corresponding to the slider’s location or entering
a numerical value of your own choosing for .

FLAT-FIELD CORRECTION

A single image prior to processing and adjustments is called a raw image. In many
cases, raw images are suitable for printing directly, perhaps after first making some
minor adjustments to brightness and contrast using the image histogram. However, a
raw image contains two significant kinds of noises that make it unsuitable for quantita-
tive purposes: First, it contains the bias signal and noise counts that increase pixel val-
ues beyond their true photometric values, which can cause significant errors in
measuring the amplitudes of object features. Second, a raw image may contain numer-
ous artifacts from the camera and microscope optical system (distortions due to detec-
tor irregularities, dust and scratches on lens surfaces, uneven illumination) that appear
as dark shadows and specks in the image and alter the true pixel values. These artifacts
are particularly visible in images with bright, uniform backgrounds (phase contrast,
DIC) and in fluorescence images with medium gray or bright backgrounds. In order to
restore photometric accuracy and remove the defects, the raw image must be adjusted by
an operation known as flat-field correction (Gilliland, 1992; Howell, 1992). Only cor-
rected images are suitable for measuring light amplitudes. Although flat-field correction
is not required to display or print an image, it is very effective in removing disfiguring
faults, so the procedure is widely used for cosmetic purposes as well. As shown in Fig-
ure 15-4 and described here, the correction procedure requires three frames:

•A raw frame contains the image signal plus signals from optical defects and elec-
tronic and thermal noises.
•A flat-field frame is the image of a smooth, featureless field without the object, and,
if possible, is made at the same focus as the raw frame. Flat frames should be bright
and utilize the full dynamic range of the camera in order to minimize the noise in
the final corrected image. If both the raw and flat-field frames are dim and noisy,
the final processed frame will be very noisy indeed. This situation is particularly
relevant to fluorescence imaging. Therefore, the exposure time for a flat-field frame
might be longer than that used for producing the raw image itself. To reduce noise