I think the "x,y" coordinate for imagemaps was a quick hack that should be
upgraded; it's very limiting. Extending it to video would be even more
limiting.
Newer approaches involve having the document specify a polygon which
constitute an anchor in an image, along with associated attributes (e.g.
destination URL, REL or REV tags, that sort of disposition information.)
That way, browsers can clearly indicate which part of the image is a valid link
and which is not, where it goes, whether it should pop up in a separate window
or replace what you're watching, etc.
I'd suggest extending that to video by just adding a couple of attributes to
the polygon (starting time and ending time, in frames or whatever.)