http://en.wikipedia.org/wiki/Streak_camera

Basically, the light hits a photosensor (kind of like a solar cell), gets converted to electricity. That electrical signal goes to a cathode ray tube, like an old tv, and makes a dot on a phosphorus screen. That screen is then recorded by a ccd sensor, just like in a standard ccd sensor like in a standard camera as the phosphor will glow for a couple of seconds. I’m guessing their 500 sensors refer to 500 lines and they adjust the cathode ray tube to move at 1-trillionth of a second in order to get all the lines on the phosphor screen, just like an old tv shoots a beam across the surface of the screen. But instead the screen is connected to a ccd sensor to see where the light hits. 

As they were saying, the camera shoots only a video of a line. So the resulting still image wouldn’t make much sense as you would have 500 lines at 500 different time. That’s what they call a 2D image, in the sense that a line is one dimension and time is the other. They must tilt a mirror to get all the lines that would make up a still image. The corresponding lines at the corresponding times are put together to make up the full video