HaaRGlyph - A New Method for Anaglyphic Reversion in Stereoscopic Videos

• Felipe Maciel Rodrigues USP
• Juliano Koji Yugoshi USP
• Rudinei Goularte USP

Resumo

In spite of recent advances in the so called 3D technology, the related literature reveals that, on the one hand, there is still a need for compression in order to reduce data volume. On the other hand, there is a lack of encoding methods capable to produce reusable 3D content, easily interchangeable between different visualization modes. In this scenario, a possible approach, but yet poorly explored, is to use the anaglyphic encoding method as a means to get both: higher compression and independence of visualization modes. That way, this work proposes the HaaRGlyph, a new technique for stereoscopic video coding based on anaglyphic method. HaaRGlyph uses spatial coding, lossless and lossy, in order to achieve better compression ratios minimally interfering with the depth perception quality. At the same time, the technique preserves enough information to allow an anaglyphic image to be reverted into the related stereo pair. This helps content interchange since different visualization modes use stereo pairs as input. The conduced experiments used objective metrics (PSNR - Peak Signal-to-Noise Ratio) and subjective metrics (DSCQS - Double Stimulus Continuous Quality Scale) applied to a stereoscopic image database. The results were compared to related techniques, pointing out that HaaRGlyph achieves better results.

Palavras-chave: Visualização estereoscópica, codificação de vídeo estereoscópico, vídeo anaglífico