Reversing Anaglyph Videos Into Stereo Pairs

Abstract

There is a diversity of strategies for stereoscopic video coding, commonly known as 3D videos. Each strategy focuses on one type of 3D visualization format, which may lead to incompatibility or depth perception problems if one strategy is applied to another format. A generic approach may be based on the proper coding of the stereo pair - every visualization format knows how to deal with stereo pairs - however, to keep the stereo pair, even a coded one, demands a high data volume. This paper proposes a method for reversing an anaglyph video back into a stereo pair. This way, a coder may convert a stereo pair into an anaglyph video, reducing the data volume by half, at least. Moreover, a correspondent decoder may use the proposed reversing method to restore the stereo pair, ensuring the visualization independence of the coding method. Tests showed that both conversion and posterior reversion resulted in videos with good quality obtaining an average of 63.09% of compression and 34,52dB of PSNR.