Graphics Reference
In-Depth Information
The R-D curves for both Orbi and Interview sequences show that the proposed configura-
tion with scalable H.264/AVC performs similar to the H.264/AVC configuration, and out-
performstheMPEG-4MACbasedconfigurationatallbitrates.TheSVCconfigurationhas
not outperformed the H.264/AVC configuration due to the negligible usage of inter-layer
prediction between the base and enhancement layers and extended header data associated
with SVC. Even though, common object boundaries are present in colour and depth map
sequences, the depth image sequence has different texture information, and less number
of objects and high frequency components compared to the colour video. Consequently,
the number of inter-layer coding modes used is very small during the encoding process.
In addition, the R-D performance of the scalable H.264/AVC configuration is slightly
affected due to the overhead bits added to the SVC bit-stream to recognize enhancement
layer data. However, the flexibility of the SVC configuration in stereoscopic video cod-
ing, such as asymmetric coding support (temporal, spatial and quality scalability for depth
image sequences), single bit-stream output, and backward compatibility, facilitates end-to-
end stereoscopic video communication chain to a greater extent. The flexible macro- block
(MB)sizesandskippedMBfeaturesavailableinH.264/AVCstandardhavehelpedtoachieve
better coding performance with the H.264/AVC and scalable H.264/AVC based configura-
tions compared to the MPEG-4 MAC configuration at all bitrates. Furthermore, it can be
observed that the configurations based on H.264/AVC provide reasonable image quality at
very low overall bitrates compared to the MPEG-4 MAC configuration.
