To be strictly accurate we should refer to this technique as 'Off Chip Video Integration' but we shall refer to it here simply as Video Integration.
Normal video cameras use a CCD chip to capture photons and this chip is read out a number of times every second to produce the frames seen as a moving video image. These frames are normally displayed as a continuously moving image on a monitor screen, or may be saved to a VCR for playback later.
Video integration captures video frames on the fly, and sums each successive frame in a 32 bits/pixel Fits (Flexible Image Transport System, the astronomical standard for image files) file. The 32 bit 'depth' of this Fits file is very important as it allows approximately 16 million individual frames to be summed before any pixel saturates.
Video integration effectively synthesises long exposures by adding together many short exposures. Non-linear image processing can then reveal the dimmer parts of the image as well as the brighter parts.
The subtraction of dark frames of the same group size removes hot pixels and amp glow.
The limiting factors for this technique are the sensitivity of the video camera, the aperture of the scope and the speed of the scope.
Typically, video integration will be done either with a surveillance type video camera, (with a lux rating less than 0.02) in conjunction with a video-capture card; a fram-integrating video surveillance type camera, or with a Webcam (which has built-in video capture capabilities).
Webcams differ widely in their light sensitivity and it is only the very sensitive ones that are suitable for video integration (If the CCD chip isn't sensitive enough to 'see' the photons, then the software cannot accumulate them into an image). Long-exposure modified webcams are the most suitable webcams for this method.
The sofware that was specially written for Video Integration and which will track and sum on the fly and much more is AstroVideo from COAA