Details
-
Improvement
-
Status: Resolved
-
Major
-
Resolution: Abandoned
-
None
-
None
Description
In case anyone reads this, and do not understand the explanation just ask.
Maybe this image technique is yet another instance
where a matrix multiplication can improve performance, if we start with
an identity matrix where we remove the 1's on cells containing the cutout as the multiplication on the linearized input cutout then it translates cleanly to a matrix multiplicationmultiplication.
If the cutout is zero the operation ends here, if another value, simpy do a outer vector multiplication and assign the cutout cells via a binary matrix matrix addition.