Except that it doesn't make much sense formulated like that. Once embedded in the video, the image has the same resolution as the video.
And if the image is smaller than the video (like with black borders), you can't automatically re-extract it. However, with my technique, you can open the image obtained by capture in an editing software and measure its actual size.