IN RE: AT&T INTELLECTUAL PROPERTY II, L.P., Appellant
from the United States Patent and Trademark Office, Patent
Trial and Appeal Board in No. 95/002, 353.
CONSTANTINE L. Trela, Jr., Sidley Austin LLP, Chicago, IL,
argued for appellant. Also represented by Christopher A.
Eiswerth, Joshua John Fougere, Washington, DC.
Michael Sumner Forman, Office of the Solicitor, United States
Patent and Trademark Office, Alexandria, VA, argued for
intervenor Michelle K. Lee. Also represented by Nathan K.
Kelley, Benjamin T. Hickman, Scott Weidenfeller.
Dyk, Mayer, and Reyna, Circuit Judges.
appeals a final decision of the Patent Trial and Appeal Board
in an inter partes reexamination. AT&T argues
that the Board improperly instituted the reexamination
proceedings and erred in finding that the challenged claims
are invalid as anticipated. The Board did not exceed its
statutory authority when instituting the reexamination and
substantial evidence supports the Board's finding of
anticipation. We therefore affirm.
case concerns methods of compressing and transmitting digital
video data. To increase the efficiency of digital video
transmission, video images are subdivided into blocks, where
each block consists of a discrete number of pixels. To
compress and transmit the image, a process called transform
coding is used to analyze and transform each block's
pixel data into a set of numerical representations, called
transform coefficients. After transmittal of the transform
coefficient data, the image is reconstructed by converting
each block of transform coefficients back into a block of
methods of transform coding vary in efficiency, and some
result in higher image quality than others. Prior art methods
of transform coding taught a method of scanning the transform
coefficients called "run-length encoding, " where
the coefficients in each block are scanned in a zigzag
pattern from top left to bottom right. This method had
certain advantages, but the zig-zagging scan pattern could
make it difficult to put the coefficients back in their
correct order and reconstruct the pixel block.
address the inefficiencies of run-length encoding, the Krause
patent, U.S. Patent No. 5, 295, 203 ("Krause"),
discloses "vector coding, " which eliminates the
need for scanning the transform coefficients in any
particular order or pattern. Vector coding involves assigning
a code word to a subset of coefficients that are selected for
transmission within a block of transform coefficients, or
within portions of the block referred to as regions. In a
preferred embodiment, Krause teaches dividing the block of
coefficients into regions and applying vector coding to each
region. As the written description explains, this is intended
to avoid complexities that arise when coding an entire block
of coefficients at once. Id. col. 7 ll. 25-35, col.
7 ll. 58-68.
independent claim 1 describes vector coding, and dependent
claim 2 describes the preferred embodiment of vector coding
regions of coefficients:
1. A method for coding video transform coefficients for
communication comprising the steps of:
providing a block of transform coefficients;
generating a vector to identify locations of a group of
coefficients from said block that qualify for transmission
according to predetermined criteria;
encoding said vector to provide a vector code word for
encoding the coefficients from said group to provide
coefficient code words for transmission;
wherein said vector code word correlates the coefficient code
words to coefficient ...