If you're going to use Java, then you can have the Java Applet connect to a
program on the web server and download the image in some "scrambled" form,
descramble it, and then display it.  The image will never pass through the
browser cache, and there will be no URL that the user can simply point to
and get the image, at least not without figuring out how to "descramble" it.

But without the cooperation of the user's client machine (which perhaps you
could get somewhat with an ActiveX control) you're not going to be able to
prevent things like the Windows Print-Screen key which copies the screen to
the clipboard and thus reproduce an exact copy of whatever you're showing
the user at exactly the same quality shown in the web browser.

When it comes down to it, once your information leaves your hands and passes
to an "untrusted" location, you cannot retain control over it.  The only way
to keep control is to never release the information into "untrusted" hands,
which means that you must use trusted *hardware* to present to information
to the end user.  Even then you can't stop him from using a camera to take a
picture of the screen, or a microphone to record what comes out of the
speakers, etc.  This is why all copy-protection schemes for information that
is communicated to human senses are ultimately doomed to fail, at least
until people are willing to have copy protection hardware installed in their
heads (and the world does seem to be moving in that direction).

G.