You can use pattern matching to do this. You will train your system to recognize the target. The output of the pattern matching function is the position of the match. If this position changes at all or a certain amount, then you can assume that the target has moved. You may want to compare target position to some stationary object, and not just frame position. You don't want camera movement to falsely assume a position change.
You can use the calibration functions in IMAQ to get the true distance, or do a simple scaling if you know how many pixels equal a certain movement distance of your target.
All of this would be inside a loop that also snaps a picture as often as you need it to.
