Datasets to follow up

In this context, one crucial aspect is the presence of only a small number of datasets that can be used for quality assessment in the field of FTV. More specifically ...... A list of available datasets is presented here below while a summary of the main characteristics of these datasets is in Table \ref{tab:datasets}:

• DIBR Images \cite{DIBRImage}: three multiview plus depth (MVD) sequences are considered: Book Arrival (1024x768 pixels, 16 cameras with 6.5 cm spacing), Lovebird1 (1024x768 pixels, 12 cameras with 3.5 cm spacing) and Newspaper (1024x768 pixels, 9 cameras with 5 cm spacing). Seven DIBR algorithms are used to create for each sequence four new viewpoints. From the created sequences, key frames are extracted and included in the dataset. Absolute Category Rating (ACR) and pair comparison have been used for Mean Opinion Score (MOS) collection.

• DIBR Videos \cite{DIBRVideo}: 102 video sequences of length 6s with 1024x768 pixel resolution frame rate between 15 and 30 frames per second are created. The original sequences are three multiview plus depth videos processed with 7 DIBR algorithms to generate 4 new viewpoints for each sequence. ACR-HR has been used for MOS collection;

• MCL 3D Database \cite{Song_2014}: this database contains 693 stereoscopic image pairs. Nine image-plus-depth sources are first selected, and a DIBR technique is used to render stereoscopic image pairs. Distortions applied to either the texture image or the depth image before stereoscopic image rendering include: Gaussian blur, additive white noise, downsampling blur, JPEG and JPEG-2000 (JP2K) compression and transmission error;

• SIAT Synthesized Video Quality Database: 10 MVD sequences and for each sequence, 14 different texture/depth quantization combinations were used to generate the texture/depth view pairs with compression distortion;

• Free-Viewpoint synthesized videos: 264 video sequences of 100 frames (around 7 seconds @15fps) in 1024x768 and 1920x1080 pixels resolution. Individual votes and MOS scores obtained by an ACR-HR experiment are provided. Six multiview sequences are considered and, from a given MVD sequence, two different view-points at one time instant t were considered. The associated depth maps were encoded through seven depth map codecs. From the decoded depth maps, fifty intermediate viewpoints (equally separated) were generated in-between the two considered viewpoints. A sequence of 100 frames (at 10 fps) was built from the 50 intermediate virtual frames to simulate a smooth camera motion from left to right and from right to left;

• High-Quality Streamable Free-Viewpoint Video: a dense set of RGB and IR video cameras is used to record videos that are compressed into a streamable 3D video format.

\label{tab:datasets}