Huh? This fractional question makes no sense at all. Assuming that you actually have a 1-channel image, as a double precision matrix, in order to even TRY to use Cholesky on it, the image array would need to be symmetric and positive definite.
Nor would it make any sense at all to want to form a Cholesky factor for that array.
So this question is simply ridiculous as posted. Sorry, but it is. You may indeed have a valid question deeply hidden in there. But for a useful answer, you need to be far more clear. Or possibly you need to reconsider why you THINK you need to build a cholesky factor for an image.
If your question does not involve an image array (YOU DID EXPLICITLY SAY IMAGE THERE) and you jest need to know how to use chol then just type this:
doc chol
or even
help chol
The help docs already tell you far more than we will tell you in this limited space.
