Relevant Research

There has been a lot of effort on representing time-based media in a printable static image form. Representation with key frames that summarize the salient content of video has been investigated extensively, resulting in multi-layer representations such as Video Manga [2]. Using such a technique, key frames are printed on paper as a representation of video. However, in this case the motion information is lost. Printing several key frames and morphing between key frames [3] is also possible in order to obtain a video effect, but many key frames need to be printed and other binary information such as audio cannot be represented as a static image.

Another attempt at utilizing paper for video display is Video Paper [4][5]. Video Paper contains key frames from video clips and provides random access to video with barcodes. In Video Paper, the barcode stores only the links to the video data and does not attempt to store an entire video sequence on paper. Therefore, access to a server is needed to play back the actual video sequence, where in our method it is not needed.

There is also some relevant work in the video compression area, mostly in error recovery of reference frames where all the bits for reference frames are not available. In most of those cases, some part of a reference frame is available and used for reconstructing the missing parts of the reference frame [6]. In multiple description coding video schemes, there could be more than one video stream representing the same video content. If the reference frame is missing from one of the streams, the other video streams can be used for replacing the missing reference frame [7]. However, these error recovery methods reconstruct reference frames from the bit stream, where as in our method a missing reference frame is constructed solely from a picture of a printed key frame.