An apparatus for merging real and virtual images in which points in three dimensions of an actual scene are obtained and selected in a random fashion and in a number sufficient to track the movement of the three dimensional scene relative to an observer wearing the apparatus. Updating of selected points occurs responsive to movement of a video camera pair observing the actual scene. The position and orientation of the unit housing the video camera pair is utilized to properly merge a generated virtual image with the real image. Selection of points and orientation of the unit relative to a three dimensional real image are obtained in real time. Microprocessor-based apparatus is utilized to perform the method of the invention.