SVS-Vistek eco4050TR, Руководство пользователя

Страница: 64 / 142

SVS-VISTEK
Feature-Set 64
7.2.11 Shading Correction
The interactions between objects,
illumination, and the camera lens might
lead to a non-uniform flatfield in
brightness. Shading describes the non-
uniformity of brightness from one edge to
the other or center towards edge(s).
This shading can be caused by non-
uniform illumination, non-uniform camera
sensitivity, vignetting of the lens, or even
dirt and dust on glass surfaces (lens).
Shading correction is a procedure to create
a flatfield image out of a non-uniform
image regardless of the reasons of the non-
uniformity. Before doing shading
correction, make sure your lens is clean
and in perfect condition. If the lens is not
clean or the lighting not uniform, the
algorithm tries to compensate these as well
– resulting in a wrong shading table and
visible artifacts, loss of details or local noise in the final image.
In theory there are several ways to correct shading:
>
In the host computer: Significant loss of dynamic range, colour
ruptures
>
In the camera, digital: better (smoother) shading than on the
computer side (10 or 12 bit), loss of dyn range
> In the camera, analog: Change gain/offset locally on sensor to get
optimum shading correction with only small changes in dynamic
range
Performing builtin shading correction
In order to perform a correction for an image with non-uniform image a
reference “white” image is captured. This will allow creating correction
values to “adjust” the pixels by individual gain settings.
8 frames are taken for averaging of white images.
Generation of the white image for correction:
The ideal white image consists of a uniform image with only one pixel
value. Pixel values lower than the brightest value are adjusted via the pixel
gain factor. The maximum gain factor is 4 (relatively to initial gain
setting). A better grey value resolution with maximum gain factor 2 can be
achieved, if the factor between the lowest and the highest pixel value of
the white image is smaller than 2.
The white image should be uniform, without saturation. To suppress small
image structures, the camera can be defocused.
The generated gain correction values are be stored to the non-volatile
memory of the camera (EPROM).
NOTICE
White balance should be completed before acquisition of
correction values for Shading Correction.
Figure 4: illustration of original
and shading corrected image