The white balance module aims to change the balance between the Red, Green and Blue channels such that a white object appears white in the acquired images. LUCID Vision Labs cameras allow for manual white balance adjustment by the user, or automatic white balance adjustment based on statistics of previously acquired frames. Different external illuminations and different sensors may render acquired images with color shift. The White Balance module allows the user to correct for the color shift by adjusting gain value of each color channel.
LUCID Vision Labs offers two types of white balance algorithm as described below. Both methods below allow for user controlled anchor points or reference points, from which multipliers are computed for each channel. The different anchor points are summarized below.
Anchors | Information |
---|---|
Min | The lowest luminance channel is used as reference while other channels are adjusted to match it. There is no chance of overflowing the pixels, however the image is darkened. |
Max | The highest luminance channel is used as reference while other channels are adjusted to match it. There is a chance of overflowing the pixels. |
Mean | The mean value of all channels is used as reference while all channels are adjusted to match the mean. There is a smaller chance of overflowing. |
Green | Green channel is used as the reference while the Red and Blue are adjusted. |
Grey World
A grey world assumes that the average of all colors in an image is a neutral grey.
White Patch
White patch has the same idea as Grey World, but only considering a section of the image (i.e. the section being the white patches). A simple way to determine such section(s) of the image is to indicate a pixel as white when R+G+B is greater than the threshold pixel value. Determining the threshold can be done using a 90% percentile of previous image. There is also a need for an additional threshold to exclude saturated pixels for better white balance adjustment.