Appeal No. 1999-2449
Application No. 08/745,584 Page 6
Gonzales et al clearly teaches at column 8, lines
11-25 that preserving the macroblock identity
simplifies significantly the derivation of motion
estimation vector data for all resolution scales
other than the highest resolution and essentially
that the motion vector data corresponding to any
resolution scale can be derived from the highest
resolution motion vector data by appropriately
scaling it down. And in an alternative way, the
full resolution motion vectors can be derived by
appropriate scaling up of lower resolution motion
vectors, and wherein additional correction (i.e.,
refinement) may be added at the higher resolution
scale (i.e. non-down sampled data) to improve the
precision of the motion vector data. As such,
Gonzales shows the same coding using the motion
vector from one search in a subsequent refinement
search and it is still the Examiner's opinion that
Gonzales et al anticipates the claimed limitations
of performing a search using down sampled data and
performing a refinement search around the offset of
the best match macroblock.
We find that Gonzales discloses (col. 6, lines 7-12) that
"[i]t is one object of this invention to preserve the identity
of MBs across a multiplicity of scales such that the overhead
is included only once, except perhaps for the refinement of
some parameters such as the accuracy of the motion vectors,"
and (col. 8, lines 19-24) that "[a]lternatively, the full
resolution motion vectors can be derived by appropriate scale
up of lower resolution motion vectors. In the latter case, an
additional correction may be added at the higher resolution
scale to improve the precision of the motion vector data."
From this disclosure of Gonzales, we agree with the
examiner that the language "additional correction may be added
Page: Previous 1 2 3 4 5 6 7 8 9 10 11 12 13 Next
Last modified: November 3, 2007