`571-272-7822
`
`Paper 20
`Entered: July 16, 2021
`
`
`
`
`
`
`
`
`
`UNITED STATES PATENT AND TRADEMARK OFFICE
`
`BEFORE THE PATENT TRIAL AND APPEAL BOARD
`
`UNIFIED PATENTS, LLC,
`Petitioner,
`v.
`ELECTRONICS AND TELECOMMUNICATIONS RESEARCH
`INSTITUTE, KWANGWOON UNIVERSITY RESEARCH INSTITUTE
`FOR INDUSTRY CORPORATION, and INDUSTRY-ACADEMIA
`COOPERATION GROUP OF SEJONG UNIVERSITY,
`Patent Owner.
`
`IPR2021-00368
`Patent 9,736,484 B2
`
`Before JAMESON LEE, SALLY C. MEDLEY, and
`NATHAN A. ENGELS, Administrative Patent Judges.
`ENGELS, Administrative Patent Judge.
`
`DECISION
`Denying Institution of Inter Partes Review
`35 U.S.C. § 314, 37 C.F.R. § 42.4
`
`
`
`
`
`
`
`
`
`IPR2021-00368
`Patent 9,736,484 B2
`
`INTRODUCTION
`I.
`Petitioner Unified Patents, LLC filed a Petition (Paper 2 (“Pet.”)) for
`inter partes review of claim 4 of U.S. Patent No. 9,736,484 B2 (Ex. 1001,
`“the ’484 patent”). Electronics and Telecommunications Research Institute,
`Kwangwoon University Research Institute for Industry Cooperation, and
`Industry-Academia Cooperation Group of Sejong University (collectively,
`“Patent Owner”), filed a Preliminary Response. Paper 10 (“Prelim. Resp.”).
`Under 35 U.S.C. § 314(a), an inter partes review may not be instituted
`unless the information presented in the Petition and any response thereto
`show “there is a reasonable likelihood that the petitioner would prevail with
`respect to at least 1 of the claims challenged in the petition.” Upon
`consideration of the Petition and the evidence of record, we determine that
`Petitioner has not demonstrated a reasonable likelihood of prevailing in
`establishing unpatentability of at least one claim of the ’484 patent.
`A. Related Matters
`The parties indicate that the ’484 patent is not the subject of any
`related administrative or judicial proceedings. See Pet. 73; Paper 3, 1.
`B. Real Parties in Interest
`Petitioner identifies itself as the real party-in-interest for Petitioner.
`Pet. 73. Patent Owner identifies itself as the real parties-in-interest for
`Patent Owner. Paper 3, 1.
`Patent Owner contends that at least Apple Inc. should have been
`identified as a real party-in-interest by Petitioner and that the Petition should
`be denied pursuant to 35 U.S.C. § 312(a) for failing to identify all real
`parties-in-interest. See Prelim. Resp. 49–56. Because we determine
`Petitioner has not demonstrated a reasonable likelihood of prevailing in
`establishing unpatentability, we do not reach this issue.
`
`2
`
`
`
`IPR2021-00368
`Patent 9,736,484 B2
`C. The ’484 Patent
`Titled, “Apparatus for Encoding and Decoding Image Using Adaptive
`DCT Coefficient Scanning Based on Pixel Similarity and Method Thereof,”
`the ’484 patent relates generally to an encoding/decoding apparatus and
`method using an adaptive Discrete Cosine Transformation (“DCT”)
`coefficient scanning based on pixel similarity. Ex. 1001, (54), 1:26–29.
`More particularly, the ’484 patent describes an encoding/decoding apparatus
`and method which performs intra prediction onto input video, predicts pixel
`similarity based on pixel similarity information of coefficients to be encoded
`that is acquired from adjacent pixels in the intra-predicted video, and
`performs scanning (e.g., DCT coefficient scanning) according to the
`predicted pixel similarity. Id. at 1:30–36.
`Copied below is Figure 5 of the ’484 patent.
`
`
`Figure 5 illustrates a typical zigzag scanning method. Id. at 5:51–52. The
`zigzag scanning method is devised in consideration that low frequency
`components of transformed coefficients acquired from the DCT and
`quantization are highly likely to be positioned in the upper left part of a two-
`dimensional plane. Id. at 5:56–60.
`Copied below is Figure 6 of the ’484 patent.
`
`3
`
`
`
`IPR2021-00368
`Patent 9,736,484 B2
`
`
`Figure 6 illustrates a typical horizontal scanning method. Id. at 5:52–54.
`The horizontal prediction mode is selected as an optimal mode when the
`pixel similarity in the horizontal direction is high. Id. at 6:16–18.
`Copied below is Figure 7 of the ’484 patent.
`
`
`Figure 7 illustrates a typical vertical scanning method. Id. at 5:54–55. The
`vertical prediction mode is selected as an optimal mode in a rate-distortion
`optimization process when the pixel similarity in the vertical direction is
`high. Id. at 6:9–11.
`Copied below is Figure 11 of the ’484 patent.
`
`
`
`4
`
`
`
`IPR2021-00368
`Patent 9,736,484 B2
`Figure 11 is a block view showing a decoding apparatus using an adaptive
`DCT coefficient scanning based on pixel similarity. Id. at 8:41–43. The
`decoding apparatus includes an entropy decoding unit 50, a scanning
`decision unit 60, and a video recovery unit 70. Id. at 8:45–48. The entropy
`decoding unit 50 receives an encoded video bitstream encoded in the
`encoding apparatus of Figure 4 using an adaptive DCT coefficient scanning
`based on pixel similarity and decodes it through an entropy decoding
`method. Id. at 8:49–52. Then, the entropy decoding unit 50 transmits the
`entropy-decoded video bitstream to the scanning decision unit 60. Id. at
`8:53–55. The scanning decision unit 60 decides a scanning method for the
`coefficients decoded in the entropy decoding unit 50 according to an intra
`prediction mode. Id. at 8:56–59. The video recovery unit 70 finally
`recovers the coefficients by using the scanning method decided in the
`scanning decision unit 60 to recover the video. Id. at 8:60–62.
`D. Challenged Claim
`Claim 4, the only claim challenged in this proceeding, is reproduced
`below with formatting and bracketed labels added for clarity.
`4. [4a] A non-transitory computer-readable storage medium storing
`instructions that, when executed by a processor, cause the processor to
`perform a method of decoding, the method comprising:
`[4b] performing entropy decoding of encoded video information in
`a bitstream to obtain transform coefficients for a current block;
`[4c] selecting a scanning mode for the transform coefficients; and
`[4d] scanning the transform coefficients based on the selected
`scanning mode;
`wherein the selecting of a scanning mode comprises:
`[4c1] selecting a horizontal scanning mode in response to the
`intra prediction mode being a vertical intra prediction mode;
`and
`
`5
`
`
`
`Reference(s)
`
`Nishi1
`Nishi
`Do,2 Kobayashi3
`Do, Kalevo4
`
`IPR2021-00368
`Patent 9,736,484 B2
`[4c2] selecting a vertical scanning mode in response to the intra
`prediction mode being a horizontal intra prediction mode.
`E. Alleged Grounds of Unpatentability
`Petitioner asserts that claim 4 is unpatentable based on the following
`grounds:
`Claim Challenged
`4
`4
`4
`4
`
`35 U.S.C. §
`102(b)
`103(a)
`103(a)
`103(a)
`
`
`Petitioner relies on the Declaration of Joseph P. Havlicek, Ph.D. (Ex.
`1002) in support of its contentions.
`II. ANALYSIS
`A. Legal Standards of Anticipation and Obviousness
`Regarding anticipation, the Federal Circuit has stated that “[a] claim is
`anticipated only if each and every element as set forth in the claim is found,
`either expressly or inherently described, in a single prior art reference.”
`Verdegaal Bros. v. Union Oil Co. of California, 814 F.2d 628, 631(Fed. Cir.
`1987). The elements must be arranged as required by the claim, but this is
`not an ipsissimis verbis test, i.e., identity of terminology is not required. In
`re Bond, 910 F.2d 831 (Fed. Cir. 1990).
`Regarding obviousness, The Supreme Court set out a framework for
`assessing obviousness under § 103 that requires consideration of four
`
`1 U.S. Patent No. 6,426,975 B1, issued July 30, 2002, (Ex. 1014, “Nishi”).
`2 English Translation of Korean Patent Application No. 0135364 B1,
`registered Jan. 13, 1998 (Ex. 1010, “Do”).
`3 U.S. Patent Application No. 2005/0281337 A1, published Dec. 22, 2005
`(Ex. 1023, “Kobayashi”).
`4 WO 2001/054416 A1, published July 26, 2001 (Ex. 1011, “Kalevo”).
`
`6
`
`
`
`IPR2021-00368
`Patent 9,736,484 B2
`factors: (1) the “level of ordinary skill in the pertinent art,” (2) the “scope
`and content of the prior art,” (3) the “differences between the prior art and
`the claims at issue,” and (4) “secondary considerations” of non-obviousness
`such as “commercial success, long felt but unsolved needs, failure of others,
`etc.” Graham v. John Deere Co, 383 U.S. 1, 17–18 (1966).5
`B. Level of Ordinary Skill in the Art
`In determining the level of skill in the art, we consider the type of
`problems encountered in the art, the prior art solutions to those problems, the
`rapidity with which innovations are made, the sophistication of the
`technology, and the educational level of active workers in the field. Custom
`Accessories, Inc. v. Jeffrey-Allan Indus., Inc., 807 F.2d 955, 962 (Fed. Cir.
`1986); Orthopedic Equip. Co. v. United States, 702 F.2d 1005, 1011 (Fed.
`Cir. 1983).
`Petitioner contends a person of ordinary skill in the art at the time of
`the invention would have had
`at least a bachelor’s degree in electrical engineering or a closely
`related scientific field, such as physics, computer science, or
`computer engineering, or similar advanced post-graduate
`education in this area, with two years of experience with video
`processing systems. A person with less education but more
`relevant practical experience, depending on the nature of that
`experience and degree of exposure to image processing devices
`and algorithms, could also qualify as a person of ordinary skill in
`the field of the ’484 patent.
`Pet. 10 (citing Ex. 1002 ¶¶ 30–31; Ex. 1001, 1:26–36, 1:56–63). Patent
`Owner agrees with Petitioner’s definition of a person of ordinary skill in the
`art. Prelim. Resp. 10. For the purposes of this decision, we apply the level
`
`
`5 The current record does not include any evidence of secondary
`considerations.
`
`7
`
`
`
`IPR2021-00368
`Patent 9,736,484 B2
`of skill advocated by Petitioner, which is supported by the testimony of
`Dr. Havlicek (Ex. 1002 ¶ 30), but delete the modifier “at least” for the level
`of education to keep the articulation from being vague and extending to the
`level of an expert.
`C. Claim Construction
`We construe claims “using the same claim construction standard that
`would be used to construe the claim in a civil action under 35 U.S.C.
`282(b).” 37 C.F.R. § 42.100(b) (2019). Specifically, we apply the
`principles set forth in Phillips v. AWH Corp., 415 F.3d 1303, 1312–17 (Fed.
`Cir. 2005) (en banc). Under that standard, the words of a claim are generally
`given their “ordinary and customary meaning,” which is the meaning the
`term would have to a person of ordinary skill at the time of the invention, in
`the context of the entire patent including the specification. Phillips, 415
`F.3d at 1312–13.
`Petitioner states that no claim terms require express construction and
`that the claims should be given their plain and ordinary meanings. Pet. 10.
`Patent Owner agrees, but Patent Owner proposes clarification of the plain
`and ordinary meaning of “horizontal intra prediction mode” and “vertical
`intra prediction mode.” Prelim. Resp. 3–9.
`Patent Owner contends the plain and ordinary meaning of the claim
`term “horizontal intra prediction mode” is a spatial prediction mode in which
`“every pixel in the current block . . . is predicted by the pixel on the same
`line and immediately to the left of the current block,” and the plain and
`ordinary meaning of “vertical intra prediction mode” is a spatial prediction
`mode in which “every pixel in the current block . . . is predicted by the pixel
`in the same column and immediately above the current block.” Prelim.
`Resp. 9–10 (citing Ex. 1002 ¶¶ 52, 54). Although Petitioner does not
`
`8
`
`
`
`IPR2021-00368
`Patent 9,736,484 B2
`directly address the customary and ordinary meaning of those terms, Patent
`Owner’s proposed constructions mirror explanations of those terms provided
`by Petitioner’s declarant Dr. Havlicek. See Ex. 1002 ¶¶ 52, 54.
`Specifically, Dr. Havlicek states “[o]ne commonly used intra
`prediction mode is called horizontal prediction. For horizontal prediction,
`every pixel in the current block . . . is predicted by the pixel on the same line
`and immediately to the left of the current block.” Id. ¶ 52. Similarly, Dr.
`Havlicek states that “[a]nother commonly used intra prediction mode is
`called vertical prediction. For vertical prediction, every pixel in the current
`block . . . is predicted by the pixel in the same column and immediately
`above the current block . . . .” Id. ¶ 54.
`The ’484 patent’s Specification describes the terms consistent with
`Dr. Havlicek’s explanation. Consistent with figures provided by Dr.
`Havlicek (Ex. 1002 ¶¶ 52, 54), Figures 2 and 3 of the ’484 patent,
`reproduced below, illustrate a pixel prediction method in the vertical
`direction and in the horizontal direction, respectively.
`
`
`
`Figures 2 and 3 of the ’484 patent depict a 4x4 block 200 having pixels
`labeled a–p. Adjacent to the top of the block 200 is a row containing
`
`
`
`9
`
`
`
`IPR2021-00368
`Patent 9,736,484 B2
`reference pixels A–H, and adjacent to the left of the block 200 is a column
`containing reference pixels I–L.
`To illustrate vertical prediction, Figure 2 depicts that “pixel a 201,
`pixel e 202, pixel i 203, and pixel m 204 are predicted based on an adjacent
`pixel A in the vertical direction” (Ex. 1001, 2:26–28), and the Specification
`further explains that pixels b, f, j, and n are predicted based on adjacent pixel
`B; pixels c, g, k, and o are predicted based on adjacent pixel C; and pixels d,
`h, l, and p are predicted based on adjacent pixel D. Id. at 2:29–34. To
`illustrate horizontal prediction, Figure 3 depicts that “pixel a 205, pixel b
`206, pixel c 207, and pixel d 208 are predicted based upon an adjacent pixel
`I in a horizontal direction,” and so forth for the remaining rows. Id. at 2:37–
`39.
`
`Both the ’484 patent and Dr. Havlicek explain that vertical intra
`prediction and horizontal intra prediction predict “every pixel in the current
`block” (Ex. 1002 ¶¶ 52, 54) using the reference pixel “immediately to the
`left” (Ex. 1002 ¶ 52) or “immediately above” (Ex. 1002 ¶ 53) the current
`block. Accord Ex. 1001, 2:24–44 (describing horizontal intra prediction and
`vertical intra prediction of each pixel in a block based on “adjacent” pixels),
`6:33–47 (describing Figure 8 as depicting horizontal intra prediction based
`on pixels “positioned adjacent to the left part of the current block” and
`vertical intra prediction based on pixels “positioned adjacent to the upper
`part of a current block”). The ’484 patent states that vertical intra prediction
`and horizontal intra prediction depicted in Figures 2 and 3 are examples of
`two of nine prediction modes provided in the H.264/Advanced Video
`Coding standard. Ex. 1001, 1:64–2:23; see Ex. 1002 ¶¶ 74–75. Notably,
`each of the nine prediction modes depicted in the H.264 standard uses
`
`10
`
`
`
`IPR2021-00368
`Patent 9,736,484 B2
`reference pixels immediately adjacent to the pixels of the current block.
`E.g., Ex. 1023, Fig. 16 (“illustrating nine intra prediction modes” (¶ 32)).
`Accordingly, the evidence of record shows that the phrases are terms
`of art. Reading the claim limitations in light of the Specification and the
`evidence of record, including Dr. Havlicek’s explanations of the phrases as
`terms of art, we agree with Patent Owner that a person of ordinary skill
`would have understood the customary and ordinary meaning of “horizontal
`intra prediction mode” is a spatial prediction mode in which “every pixel in
`the current block . . . is predicted by the pixel on the same line and
`immediately to the left of the current block” and that the customary and
`ordinary meaning of “vertical intra prediction mode” is a spatial prediction
`mode in which “every pixel in the current block . . . is predicted by the pixel
`in the same column and immediately above the current block.”
`D. Summary of Prior Art References
`1. Nishi
`Titled “Image Processing Method, Image Processing Apparatus and
`Data Recording Medium,” Nishi describes image processing methods, image
`processing apparatuses and data recording media in which, in variable-
`length coding of frequency components of an interlaced image signal, a
`sequence of the frequency components is adaptively rearranged. Ex. 1014,
`(54), 1:9–13.
`Nishi describes the conventional prior art (“a conventional image
`processing apparatus”) with respect to Figures 29–32. Figures 31(a)–31(c)
`are copied below.
`
`11
`
`
`
`IPR2021-00368
`Patent 9,736,484 B2
`
`
`Figures 31(a)–31(c) are diagrams each illustrating a scanning order in a scan
`which is selected in scan changing method. Id. at 32:12–14. As described in
`Nishi, a scan method is changed according to ON/OFF state information of
`an AC prediction in intra-frame prediction. Id. at 4:14–16. Further, when
`AC prediction is in the ON state, a scan method is changed according to a
`reference direction of prediction. Id. at 4:16–18.
`When AC prediction is in the OFF state, a scan of quantized values is
`executed in the order shown in Figure 31(a). Id. at 4:25–27. In this order,
`the quantized values are uniformly scanned in the order from low-frequency
`components to high-frequency components. Id. at 4:31–33. When AC
`prediction is performed and a vertical direction is referred, a scan of
`quantized values is executed in the order shown in Figure 31(b). Id. at 4:33–
`35. In this order, the quantized values are scanned with a priority given to
`horizontal direction. Id. at 4:39–41. When AC prediction is performed and
`a horizontal direction is referred, a scan of quantized values is executed in
`the order shown in Figure 31(c). Id. at 4:41–44. In this order, the quantized
`values are scanned with a priority given to a vertical direction. Id. at 4:47–
`49.
`
`12
`
`
`
`IPR2021-00368
`Patent 9,736,484 B2
`2. Do
`Titled “Method and Apparatus for Encoding DCT Blocks Using
`Block-Adapting Scan,” Do describes a quantization scanning technique for a
`DCT coefficient in an image encoding and decoding process, and, in
`particular, a method and apparatus for encoding DCT blocks using a block-
`adaptive scan to increase the efficiency of a variable-length coding by
`adaptively transforming a scan pattern. Ex. 1010, 8–2.
`Figure 4 is copied below.
`
`
`Figure 4 illustrates a set of five scan patterns. Id. at 8–3. The sets of scan
`patterns are stored in the scan pattern storage units installed in the encoder
`and the decoder, respectively, and when the scan pattern is determined in the
`scan pattern determination units, the processor refers to it. Id. More
`specifically, Do’s method and apparatus use a motion compensation image
`to determine a suitable scan pattern from a set of scan patterns for each
`
`13
`
`
`
`IPR2021-00368
`Patent 9,736,484 B2
`DCT-converted block of the motion compensation prediction difference
`image, converts the block of the motion compensated image into DCT, and
`analyzes and uses the coefficient for efficient and simple extraction of
`contour features. Id.
`3. Kobayashi
`Titled “Moving Image Coding Apparatus,” Kobayashi describes an
`image coding apparatus that determines an image pattern of image data and,
`based on the determined image pattern, selects a prediction mode for
`generating predicted pixel values by predicting pixel values in a frame using
`pixel values in the same frame. Ex. 1023, (54), (57).
`Figure 16 is copied below.
`
`14
`
`
`
`IPR2021-00368
`Patent 9,736,484 B2
`
`
`
`Figure 16 shows directions of prediction for use in nine different intra
`prediction modes. Id. at ¶ 114. For example, in the case of the vertical intra
`prediction mode (Mode 0), values of the pixels adjacent to the target block
`in the vertical direction are used as prediction values, and it is predicted that
`these pixel values continue in the vertical direction. Id. In addition to the
`vertical intra prediction mode (Mode 0), eight intra prediction modes (Mode
`1 to Mode 8) are provided including a horizontal intra prediction mode
`(Mode 1). Id. at ¶ 115.
`Figure 1 is copied below.
`
`15
`
`
`
`IPR2021-00368
`Patent 9,736,484 B2
`
`
`Figure 1 is a block diagram illustrating processing for intra prediction. Id. at
`¶ 46. The intra prediction unit 609 includes an image pattern determination
`unit 101, an intra prediction mode designation unit 102, a selector 103, a
`vertical intra prediction unit 104, a horizontal intra prediction unit 105, a DC
`intra prediction unit 106, and a selector 107. Id. at ¶ 47. The image pattern
`determination unit 101 determines an image pattern by performing a
`Hadamard transform on an input image. Id. The intra prediction mode
`designation unit 102 designates an optimal intra prediction mode from
`among a plurality of intra prediction modes based on the image pattern
`determined by the image pattern determination unit 101. Id. The selectors
`103 and 107 select one of the vertical intra prediction unit 104, the
`
`16
`
`
`
`IPR2021-00368
`Patent 9,736,484 B2
`horizontal intra prediction unit 105 and the DC intra prediction unit 106,
`corresponding to the intra prediction mode designated by the intra prediction
`mode designation unit 102. Id. The vertical intra prediction unit 104
`performs intra prediction using a vertical intra prediction mode. Id. The
`horizontal intra prediction unit 105 performs intra prediction using a
`horizontal intra prediction mode. Id. The DC intra prediction unit 106
`performs intra prediction using a DC intra prediction mode. Id.
`4. Kalevo
`Titled “A Method for Encoding Images, and an Image Coder,” Kalevo
`describes a method for encoding a digital image in which the digital image is
`divided into blocks. Ex. 1011, (54), (57). In the method, a spatial prediction
`for a block is performed to reduce the amount of information to be
`transmitted, where at least one prediction method is defined. Id. at (57).
`Figures 5a–5l are copied below.
`
`
`
`17
`
`
`
`IPR2021-00368
`Patent 9,736,484 B2
`
`
`Figures 5a–5l depict prediction methods utilized by the method for encoding
`a digital image. Id. at 17:30–32. According to an embodiment of the
`method, a subset of prediction methods for each context class combination is
`defined and the prediction methods are prioritized (ranked) in each subset.
`Id. at 18:10–13. Then the prediction method used to predict the content of
`the current block C is selected from a subset of prediction methods. Id. at
`18:13–15. Prediction methods P5 (Figure 5e) and P9 (Figure 5i) predict
`vertical and horizontal shapes in the current block C by extending image
`details into the current block C, either from above or from the left. Id. at
`22:17–19. Depending on the selected method (P5 or P9), the reference pixel
`values at the boundary of either block U or L are copied to the current block
`C as depicted in Figures 5e and 5i. Id. at 22:19–22.
`E. Grounds 1 and 2
`With a limitation-by-limitation comparison of Nishi to claim 4,
`Petitioner argues claim 4 is unpatentable as anticipated by Nishi (Ground 1)
`or obvious in view of Nishi (Ground 2). With reference to the bracketed
`
`18
`
`
`
`IPR2021-00368
`Patent 9,736,484 B2
`labels used in the reproduction of claim 4 above, we address limitations
`[4c1] and [4c2], which we determine to be dispositive of Grounds 1 and 2.
`[4c] selecting a scanning mode for the transform coefficients; and
`wherein the selecting of a scanning mode comprises:
`[4c1] selecting a horizontal scanning mode in response to the
`intra prediction mode being a vertical intra prediction mode;
`and
`[4c2] selecting a vertical scanning mode in response to the
`intra prediction mode being a horizontal intra prediction
`mode.
`Petitioner argues Nishi discloses directional intra prediction, and
`Petitioner provides the following annotated version of Nishi’s Figure 30 to
`support its arguments. Pet. 15.
`
`
`
`Pet. 15. Petitioner’s annotated version of Nishi’s Figure 30 depicts a
`macroblock that has 16x16 pixels divided into four sub-blocks of 8x8 pixels.
`Pet. 16. The block to be encoded (the “current block”) is labeled X and
`shaded yellow, with reference block R1 above the current block and
`reference block R2 left of the current block. Pet. 16. AC coefficients for R1
`are in the uppermost row, shaded green, and AC coefficients for R2 are in
`
`19
`
`
`
`IPR2021-00368
`Patent 9,736,484 B2
`the left most column, shaded red. Pet. 16. According to Petitioner, “Nishi
`explains that either horizontal (lateral) or vertical (longitudinal) intra
`prediction is performed depending on whether block R1 or R2 is to be
`referenced.” Pet. 16.
`Contrary to Petitioner’s arguments, however, Nishi does not disclose
`either horizontal intra prediction or vertical intra prediction consistent with
`the claims of the’484 patent. As explained above, we determine based on
`the evidence of record that the meaning of “horizontal intra prediction
`mode” is a spatial prediction mode in which “every pixel in the current block
`. . . is predicted by the pixel on the same line and immediately to the left of
`the current block” and that the meaning of “vertical intra prediction mode” is
`a spatial prediction mode in which “every pixel in the current block . . . is
`predicted by the pixel in the same column and immediately above the
`current block.”
`Dr. Havlicek states that Nishi’s Figure 30 “shows an illustration of
`how a current block (‘X’) can be intra predicted using values in adjacent
`blocks” (Ex. 1002 ¶ 154), but Dr. Havlicek does not state, much less explain,
`how Nishi’s discussion of using AC coefficients from the uppermost row or
`leftmost column of an adjacent block satisfies claim 4’s recitation of vertical
`intra prediction mode or horizontal intra prediction mode. Nothing in Nishi
`discloses predicting every pixel in the current block using an immediately
`adjacent reference pixel. See Prelim. Resp. 11 (arguing Nishi’s AC
`prediction “is a transform coefficient (NOT spatial) prediction mode in
`which AC coefficients (NOT pixels) of only one (NOT every) row or column
`is predicted by a transform coefficient (NOT pixel) of a row or a column on
`a distant (NOT immediately adjacent) boundary of a neighboring block”).
`
`20
`
`
`
`IPR2021-00368
`Patent 9,736,484 B2
`Accordingly, we determine Petitioner has not satisfied its burden to
`show a reasonable likelihood that it would prevail in establishing that claim
`4 is anticipated by Nishi. Petitioner does not adequately articulate an
`obviousness argument to address these deficiencies of Nishi, and we
`determine that Petitioner has not satisfied its burden to show a reasonable
`likelihood that it would prevail in establishing that claim 4 is unpatentable as
`obvious in view of Nishi.
`F. Grounds 3 and 4
`With a limitation-by-limitation comparison of claim 4 to Do and
`Kobayashi (Ground 3) and Do and Kalevo (Ground 4), Petitioner argues
`claim 4 is unpatentable as obvious. For the reasons explained below, we
`determine Petitioner has not articulated an adequate rationale for combining
`the references’ teachings for each ground to arrive at the claimed invention,
`which is dispositive for Grounds 3 and 4.
`Petitioner acknowledges that Do teaches applying a scan pattern to a
`“quantized DCT block of the motion compensation prediction difference
`image.” Pet. 39 (quoting Ex. 1010 p. 8-3). As noted by Patent Owner and
`also Petitioner’s expert Dr. Havlicek, Do’s reference to motion
`compensation prediction indicates that Do teaches inter frame prediction, not
`intra prediction. Ex. 1002 ¶ 56 (noting that predicting a group of pixels in
`the current frame by using pixel values from a different location in a
`different frame is called motion compensation prediction), ¶ 174 (“Do does
`not mention intra prediction modes, including a horizontal or vertical intra
`prediction mode”); Prelim. Resp. 18–20; see Ex. 1010 p. 8-3 (“the basic idea
`of the present invention is that the contour features of the original image
`captured by the camera, the motion compensated image, and the motion
`compensation prediction image are similar to each other”).
`
`21
`
`
`
`IPR2021-00368
`Patent 9,736,484 B2
`Petitioner addresses the fact that Do does not teach intra prediction by
`noting that Do’s filing date “predated the adoption of directional intra-
`prediction modes in common standardized video codecs.” Pet. 40 (citing
`Ex. 1010 p. 8-1). Petitioner argues that “[i]n the years after the filing of Do,
`standardized video codecs adopted a scheme of using various directional
`intra prediction modes to perform intra prediction on a current block using
`values from reference blocks to achieve additional compression.” Pet. 40.
`Petitioner suggests multiple different reasons that a person of ordinary
`skill would have added directional intra prediction to Do. First, Petitioner
`argues that a person of ordinary skill would have “looked to add directional
`intra prediction to the system of Do to provide the same advantages that
`directional intra prediction has provided to more modern video coding
`schemes such as H.264, such as increasing image compression.” Pet. 43; see
`Pet. 60–61. Further, Petitioner contends that the fact that Do was filed
`before directional intra prediction had been incorporated into the most
`ubiquitous video encoding standards, shows “it was ready for improvement
`using the known directional intra prediction techniques.” Pet. 43 (citing Ex.
`1002 ¶¶ 186–188, 192; Ex. 1023 ¶¶ 12–13). Petitioner also suggests a
`person of ordinary skill would have looked at the intra prediction encoding
`mode as a “substitute for calculating the directional contours within an
`image as in Do based on the combined teachings of Do and Kobayashi,”
`suggesting that such a change would have been substitution of one known
`method for another yielding predictable results. Pet. 44, 46, 63–64
`(addressing Do in combination with Kalevo). Petitioner contends that such a
`substitution would have reduced the computational complexity of Do’s
`calculations. Pet. 44–45, 47, 62–63, 65. And Petitioner also contends that
`“[o]nce directional intra prediction is added to Do’s coding scheme, the
`
`22
`
`
`
`IPR2021-00368
`Patent 9,736,484 B2
`[person of ordinary skill] would have understood from Kobayashi’s
`teachings that the intra prediction method selected for coding a particular
`block would have acted as a quickly-determinable proxy for calculating an
`edge/contour direction according to the teachings of Do.” Pet. 43–44 (citing
`Ex. 1002 ¶¶ 187, 191); see Pet. 61–62 (addressing Do in combination with
`Kalevo).
`Patent Owner argues, and we agree, that Petitioner does not
`adequately address the significant differences between Do’s teachings,
`including inter prediction, and the claimed invention, and we determine that
`Petitioner also does not adequately explain how and why a person of
`ordinary skill would have combined Do’s teachings with Kobayashi or
`Kalevo to arrive at the claimed invention without improper hindsight.
`Prelim. Resp. 18–20, 22–24. First, although Petitioner acknowledges that
`Do does not teach or even mention intra prediction, Petitioner does not
`directly address the fact that Do teaches inter prediction. See Pet. 39–40.
`That leads to Petitioner freely referring to adding intra prediction to Do
`without explaining whether intra prediction can be added to Do without
`fundamentally changing the operation of Do. It is not explained by
`Petitioner whether the inter prediction of Do can even coexist with the
`addition of intra prediction, and if so, how. On the record presented by
`Petitioner, we do not know what is meant by adding intra prediction to Do.
`For instance, does the resulting system, subsequent to the proposed addition,
`perform both Do’s inter prediction and the additional intra prediction? If so,
`the record does not show how that would work.
`Further, contrary to Petitioner’s arguments, the mere fact that Do was
`filed before directional intra prediction became part of some standard coding
`scheme does not establish that Do was ready for improvement using
`
`23
`
`
`
`IPR2021-00368
`Patent 9,736,484 B2
`directional intra prediction or that such a modification would have been
`obvious. Do describes taking advantage of temporal redundancies, while the
`claimed invention’s intra frame prediction centers on spatial redundancies.
`See Ex. 1002 ¶¶ 49–50, 56 (noting the differences between temporal
`prediction and spatial prediction). Dr. Havlicek’s testimony downplays the
`relevance of motion compensated prediction to t