Base-Anchored Motion (BAM)
In the previous two chapters, we have presented the BIHA and the FOHA scheme for highly scalable video compression. Both these schemes employ reference-based hierarchical motion field anchorings; that is, at each level of the temporal hierarchy, motion fields are predictively coded. For this, motion fields from coarse levels are mapped to the same or finer temporal levels in order to predict motion information. We have seen that in FOHA, the geometrical consistency of the mapped motion fields can be guaranteed. However, there is no way of ensuring that multiple interpolated frames are interpolated in a temporally consistent way. Furthermore, as observed at the end of Sect. 6.5, the problem is that the mapped motion information in general does not fall onto integer grid locations, and hence rounding errors are inevitable.
- 1.D. Rüfenacht, D. Taubman, Temporally consistent high frame- rate upsampling with motion sparsification, in IEEE International Workshop on Multimedia Signal Processing (2016) (cited on pages vii, 143)Google Scholar
- 2.L. Xu, J. Jia, Y. Matsushita, Motion detail preserving optical flow estimation. IEEE Trans. Pattern Anal. Mach. Intell. 1744–1757 (2012) (cited on pages 43, 44, 49, 73, 95, 102, 128, 130, 131, 134, 137, 139, 159, 185–187)Google Scholar