Image Processing Reference
In-Depth Information
Fig. 5.12 The Micro-image Refilling step to generate a IL reference picture [ 15 ] (Copyright
©
2013, IEEE)
Fig. 5.13 Example of a SS
reference picture (formed
by the previously coded and
reconstructed area of the
current frame itself)
5.4.2 The Self-Similarity Compensated Prediction
A scheme for self-similarity estimation and compensation was firstly proposed in
[ 24 ], in order to improve the performance of H.264/AVC [ 17 ] for 3D holoscopic
video coding. In this scheme, the inherent spatial cross-correlation between neigh-
boring micro-images is seen as type of spatial redundancy, referred to as self-
similarity, which is then exploited to improve the coding efficiency. More recently,
in [ 25 ], the authors proposed to introduce the self-similarity compensated predic-
tion into HEVC so as to take advantage of the flexible partition patterns used in this
video coding standard. A similar scheme, referred to as Intra Motion Compensation
[ 26 ], has been proposed in the literature for screen content coding (text and graphic
regions) in the HEVC range extension developments [ 27 ].
Similarly to motion estimation, the self-similarity estimation process uses block-
based matching in order to find a best match—in terms of a suitable matching
criterion—for prediction of a block area being encoded. However, in this case, the
allowed search area is the previously coded and reconstructed area of the current
picture, as illustrated in Fig. 5.13 . This previously coded and reconstructed area of
the current frame itself forms a new reference picture, referred to as SS reference.
Search WWH ::




Custom Search