|
Wavelet-based Highly Scalable Video Coding
Overview
Even though there have been significant improvements in
the field of scalable video coding, the rate-distortion performance
of scalable video coders remains inferior to single-layer
coding. One reason for this is the fact that scalable video
coders have problems at discontinuities in the motion field;
band-limited sampling of motion fields smooths out the
sharp transitions at moving object boundaries, which creates
non-physical motion – a value interpolated between the motion
of the foreground and the background object is in general
a poor predictor.
We propose to use ‘‘physical’’ motion, from which we estimate motion discontinuity information;
such discontinuity information allows
the inference of auxiliary information, which has to be explicitly
coded as side information in standard video codecs. More
surprisingly, in combination with ‘‘true’’ motion fields, it allows
to flip around the anchoring of motion fields, which proves
to yield geometrically consistent predictions with quantized
motion; in addition, significant savings in terms of motion
field coding cost can be achieved. We refer to this scheme as
bidirectional hierarchical anchoring of motion fields.
While particularly useful for highly scalable video coding, many
of the properties of the proposed scheme are also beneficial
for single-layer coding.
Related Publications and Links to Project Pages
Journal:
 |
D. Ruefenacht, R. Mathew, and D. Taubman, “A Novel Motion Field Anchoring Paradigm for Highly Scalable Wavelet-based Video Coding,” IEEE Transactions on Image Processing (TIP).
Preprint | DOI | Project Webpage
|
Conference:
 |
D. Ruefenacht, R. Mathew, and D. Taubman, “Hierarchical Anchoring of Motion Fields for Fully Scalable Video Coding,” IEEE International Conference on Image Processing (ICIP), Paris, France, 2014.
Full text | Poster
|
 |
Top 10% Award
D. Ruefenacht, R. Mathew, and D. Taubman, “Bidirectional Hierarchical Anchoring of Motion Fields for Scalable Video Coding,” IEEE International Workshop on Multimedia Signal Processing (MMSP), Jakarta, Indonesia, 2014.
Full text | DOI | Presentation Slides | Poster
|
 |
D. Ruefenacht, R. Mathew, and D. Taubman, “Motion Blur Modelling for Hierarchically Anchored Motion with Discontinuities,” IEEE International Workshop on Multimedia Signal Processing (MMSP), Xiamen, China, 2015.
Preprint | Poster | Project Webpage
|
|
