So apparently part of the fallout from the Controversial Content study is a recommendation "that the Wikimedia Foundation develop a feature to allow Wikimedia project users to opt into a system that would allow them to easily hide classes of images from their own view"
Rather than developing a significant new feature I would suggest using Adblock Plus
http://en.wikipedia.org/wiki/Adblock_Plus
Adblock Plus is under the MPL so a stripped down or streamlined version could be produced if required. Adblock Plus already makes it easy to block predefined sets of images. For example if a user were to dislike pictures of giant isopods they could download the filter that blocked all the giant isopod images. If these filters were to be stored localy we would probably need a new namespace or perhaps project. In the case of the giant isopod filter the content of the filter would look something like:
[Adblock] ! Checksum: l4nEGmZz7f1kh8Pfszr2rg *Giant_isopod.jpg* *Bathynomus_giganteus.jpg* *Bathynomus_giganteus_NOAA.jpg* *Front_View_Isopod_West_Sirius_Rig_GOM.JPG* *Isopod_from_West_Sirius_Rig_GOM.JPG*
although the average user should never see that.
There are a number of advantages to this approach: 1)It is entirely under the users control. They can chose what filters to install, edit any filters they install and turn them off on a page by page basis 2)It doesn't require logging in. So it doesn't suddenly stop working when someone is logged out for whatever reason 3)It allows for a potentially unlimited number of filtering lists 4)The lists don't have to be hosted on wikipedia allowing groups to set up common filters without having to work through us 5)This literally works right now. You could drop that list into an Adblock Plus install and see no giant isopods.
The downside: 1)We would need to check how compatible the filters are with equivalent software on chrome and opera 2)There is no equivalent for internet explorer 3)Phones could be a problem although if there is actual demand I assume someone will produce an app.