User manual

Small Vision System User Manual 20
2.1.12 Color
As of Version 2.1, SVS supports color input and display. Besides the two monochrome left/right
stereo channels, there is a third color channel that corresponds to the left image, with images in RGB 32-bit
format, and optionally a fourth color channel for the right image. The color channels do not participate in
stereo processing, but can be useful in applications that combine color and stereo information, for example,
object tracking. Usually only the left color channel is needed, since the left image is the reference image
for stereo disparities and 3D information.
Color information from the MEGA-D digital head (STH-MD1-C) is input as raw colorized pixels, and
converted by the interface library into two monochrome and one or two RGB color channels. The main
color channel corresponds to the left image, which is the reference image for stereo. The color image can
be de-warped, just like the monochrome image, to take into account lens distortion (Section 4). Optionally,
a second color channel is available for the right image.
Color information from the camera is input only if the Color button is pressed on the main window
(Figure 2-1). To get the right color image, use the SetColor() command from an application.
Because the typical color camera uses a colorizing filter on top of its pixels, the color information is
sampled at a lower resolution than a similar non-colorized camera samples monochrome information. In
general, a color camera has about ¼ the spatial resolution of a similar monochrome camera. To
compensate for the reduced resolution, use binning (Section 2.1.6) to increase the fidelity of the image.
For example, if you need a 320x240 frame size, use 640x480 and binning x2.
The relative amounts of the three colors, red/green/blue, affects the appearance of the color image.
Many color CCD imagers have attached processors that automatically balance the offsets among these
colors, to produce an image that is overall neutral (called white balance). The MEGA-D provides manual
color balance by allowing variable gain on the red and blue pixels, relative to the green pixels. Manual
balance is useful in many machine vision applications, because automatic white balance continuously
changes the relative amount of color in the image.
The manual gain on red and blue pixels is adjusted using the Video Parameters window (Section
2.1.9). For a particular lighting source, try adjusting the gains until a white area in the scene looks white,
without any color bias.