🚀 go-pugleaf

daniel@marl2.marl.panasonic.com (Daniel C. Gnanaprakasam) writes:

>I am looking for literature on computing gradient of RGB images.  Most
>of the standard image processing books define gradient for monochrome images.
>Ideally the operator should mark edges between regions of different color,
>but the regions themselves could be isoluminous if only the Y component
>of the RGB is looked at.

The gradient operator is defined on scalar fields (images).
The components of the gradient vector are the partial derivatives
of the scalar with respect to the spatial variables.
For vector fields you have a matrix of partial derivatives.
For a three component images, there will be six of them.
If all you want to do is segment regions, then you could treat the
six partial derivatives as the components of a 6D vector and
threshold on the length of this vector.  Compute the gradients of
the component r g and b images, and then possibly transform
the values to normalize the different color directions.

-- 

	Jeff Mulligan (jbm@vision.arc.nasa.gov)
	NASA/Ames Research Ctr., MS 262-2, Moffett Field CA, 94035-1000
	(415) 604-3745