Thanks for your reply, To clarify,

My endeavours are to build a system that is capable of calibrating
video to a traceable reference.

I was interested in the maths behind the problem. A noteable
suggestion has been bought to my attention. This is for PAL however it
may be suitable for NTSC also.

Sample an RGB value of a pixel. From this determine YUV
After Gamma Correction RGB become R'B'G'

RGB were first normalised (I'm assuming that the capture card is
already gamma corrected ???)

R' = R captured/255
G' = G captured/255
B' = B captured/255

Y = 0.299R' + 0.587G' + 0.114B'

U = -0.147R' - 0.289G' + 0.436B'

V = 0.615R' - 0.515G' - 0.100B'

Also note:
U = 0.492  ( B' - Y )
V = 0.0877 ( R' - Y)

The Vector phase can magnitude are obtained by converting U & V from
cartesian to polar form.

The vector phase compares very accurately ( < 1% error) of that
measured with a calibrated vector scope.

The magnitude required scaling and produced 5% error.

During investigation of the error, it was noted that the PCI-1411
appeared to load the chrominance component slightly while the
luminance remained intact?

Although the above solution 'seems' to work well. It did require an
arbitary scaling factor on the magnitude. It would be interesting to
hear thoughts on this science of video measurement with to particular
reference to the NI capture cards.

References:
Video Demystified ISBN 1-878707-09-4)
http://www.fourcc.org/fccyvrgb.php

Reply via email to