Integral images compression scheme based on view extraction

Antoine Dricot (1, 2, 3) , Joel Jung (1) , Marco Cagnazzo (2, 3) , Beatrice Pesquet-Popescu (2, 3) , Frederic Dufaux (2, 3)

1 Orange Labs [Lannion]
2 MM - Multimédia
3 TSI - Département Traitement du Signal et des Images

Integral imaging is a glasses-free 3D video technology that captures a light-field representation of a scene. This representation eliminates many of the limitations of current stereoscopic and autostereoscopic techniques. However, integral images have a large resolution and a structure based on microimages which is challenging to encode. In this paper a compression scheme for integral images based on view extraction is proposed. Average BD-rate gains of 15.7% and up to 31.3% are reported over HEVC. Parameters of the proposed coding scheme can take a large range of values. Results are first provided with an exhaustive search of the best configuration. Then an RD criterion is proposed to avoid exhaustive search methods, saving runtime while preserving the gains. Finally, additional runtime savings are reported by exploring how the different parameters interact.

Domains: Computer Science [cs]; Signal and Image Processing; Engineering Sciences [physics]; Signal and Image processing

Antoine Dricot, Joel Jung, Marco Cagnazzo, Beatrice Pesquet-Popescu, Frederic Dufaux. Integral images compression scheme based on view extraction. European Signal Processing Conference (EUSIPCO 2015), Sep 2015, Nice, France. ⟨hal-01433758⟩