Help with using vision processing software with Java API

[criogod]

My team is in its second year and this will be the first year we are incorporating vision processing in our code for the robot. We used Java our rookie season and I am familiar with most aspects of the Java API. What I want to know is the basic steps to use a vision processing software (ie RoboRealm, OpenCV, SimpleCV), in conjuction with the Java API.

Here are my main questions:

1. Is the vision processing software running on the driver station pc or the cRIO?
2. How do you utilize data from the vision software in the actual robot code?
3. If the vision software is on the cRIO, do we need a special version of the software compatible with the cRIO and/or will it take up a lot of processing power?
4. Is it possible to use the axis camera over the network to provide a target tracking system viewable from the driver station during teleop?

Thank you so much in advance for helping me out with these issues, I am looking forward to learning a lot of cool things this season!