The transmission aspects of 3D video - 3D Video Processing and Transmission - page 101

Video over IP often suffers from packet losses. The congestion and noise/interference/fad-
ing are the maincausesforpacketlossesinwiredandwirelesslinksrespectively.Therefore,
effectiveerrorcorrection approaches (e.g. Automatic Repeat-reQuest (ARQ), Forward Er-
ror Correction (FEC)), joint source channel coding techniques and error concealment
techniques are necessary to send 3D services over IP networks more effectively. The
performance of 3D video transmission over wireless channels which is considered as
bandwidth limited and error prone, can be further improved with Joint Source Channel
Coding (JSCC) approach which is an effective method to overcome such challenges [77].
The study carried out in [78] proposes a JSCC scheme for colour plus depth stereoscopic
video. The advanced channel coding approaches for 3D video are addressed in several
studies [79]. For example, in [80], stereoscopic video streaming using FEC techniques are
investigated. The left and right image frames are classified into layers based on their con-
tribution towards final reconstructed quality. Then the layers are transmitted based on the
principalofUnequalErrorProtection(UEP)usingdifferenterrorcorrectioncodes.Further-
more,MultipleDescriptionCoding(MDC)approachesareproposedforstereoscopicvideo
whichallowssendingvideowithacceptablequalityboundsatbadchannelconditions [81].
However, the use of error correction codes with 3D video is somewhat unjustifiable due to
the high demand for bandwidthby3Dvideocontentitself.Therefore,thisresearchexplores
differentperspectivestoprotect3D video over networks without sending additional data.
For example, transmission power can be allocated unequally for 3D video components de-
pending on their contribution towards perceptual quality. In a way the adapted approaches
exploit cross-layer design using the perceptual aspects of 3D video.
Error concealment is necessary to perform at the decoder in order to reduce the temporal
error propagation caused by unpreventable packet losses. The conventional error conceal-
ment tools for 2D video can be adapted in recovering the errors of corrupted 3D video
[82][83]. Error concealment algorithms have been proposed for left and right view based
stereoscopic video utilizing the additional data from the corresponding image sequence
[84] [85]. Bilen et al propose two methods for full frame loss concealment in left and
right based stereoscopic video using overlapped block motion and disparity compensation
[86]. Moreover, in [87] an approach to recover the entire right frame in stereoscopic video
transmission is described based on the relativity of prediction modes for right frames.
However, the above mentioned studies are aimed at providing error resilience for left and
right view based stereoscopic video. Therefore, error concealment of colour plus depth
stereoscopic video is considered in this topic. Three error concealment methods are pro-