My advice would be to not attempt to force a redirect on anyone viewing images. If they click the image itself (which I do almost by reflex), they'll head directly through to the web site anyway; there's also the dedicated Visit page and View original image buttons.
By not immediately loading the page underneath the image, Google's actually reducing your server load and bandwidth usage. Why would you want to regain all this unnecessary extra traffic? By not having it hitting your server at random, you're actually now seeing much more realistic traffic volumes - and your visitor metrics will be vastly more accurate as a result.
NB: where a page would load under the image on the previous version of Google Images, you should NOT count this as an organic visit. In fact, it's barely even classifiable as a passive visit - most users will view the image, click on the direct link in the right-hand bar then close the first tab altogether. They've not even viewed the page properly yet you've had to pay for the resources to serve the content...
This is actually a boon for people looking to optimise SEO and SEF because now they don't have to focus on working out what's just Google Images spider traffic and what's legitimate clickthroughs from humans.
Although I've not tested, I believe Google would either disregard a subsequent htaccess redirect (just saving the first image and closing the HTTP connection) or just not spider the image whatsoever, so you may end up doing more harm than good to your rankings.
Judging by the behaviour of web sites where referrer-based redirects are already in place (to immediately redirect someone to a web page if they click through to an image URL without providing an on-domain referrer), this kind of insistence on denying people access to the images or forcing them to go via a web page hasn't been effective for some time because people can always just lift the Google cached copy.
