If it loads in the browser, the it isn't secure.
However, if a Java applet or an ActiveX control loads, then it may be possible to secure the image. e.g. If the browser window loses focus, blackout the image. And on any mousedown or keydown event, blackout the image.
BUT, even then you may have a situation where the user is running a virtual PC or virtual OS and you can't capture the mouse or key events. In that case, security is lost again because you can't know anything about the client because the client is a client of a larger client.
Examples would be "Virtual PC", PC Anywhere, and Terminal Services.
In order to overcome that problem, your Java or ActiveX control needs to be extremely intrusive into the client, and many people would consider it a violation of privacy and a serious security risk. The average user wouldn't know, but eventually the control would get bad press and reviews.
I'm not even sure that a control would run properly to protect against the virtual client scenario because of the massive degree to which you would need to "invade" the client. Built in security may prevent that.
Does this make sense to anyone?