`571-272-7822
`
`Paper 11
`Date: October 13, 2020
`
`UNITED STATES PATENT AND TRADEMARK OFFICE
`
`BEFORE THE PATENT TRIAL AND APPEAL BOARD
`
`GOOGLE LLC,
`Petitioner,
`v.
`UNILOC 2017 LLC,
`Patent Owner.
`
`IPR2020-00757
`Patent 7,012,960 B2
`
`
`
`
`
`
`
`
`
`Before SALLY C. MEDLEY, MICHAEL R. ZECHER, and
`NABEEL U. KHAN, Administrative Patent Judges.
`KHAN, Administrative Patent Judge.
`
`DECISION
`Denying Institution of Inter Partes Review
`35 U.S.C. § 314
`
`
`
`
`
`
`
`
`
`IPR2020-00757
`Patent 7,012,960 B2
`
`I.
`
`INTRODUCTION
`Background and Summary
`A.
`Google LLC (“Petitioner”) filed a Petition (Paper 1, “Pet.”) requesting
`an inter partes review of claims 1, 4, and 5 (“the challenged claims”) of U.S.
`Patent No. 7,012,960 B2 (Ex. 1001, “the ’960 Patent”). UNILOC 2017 LLC
`(“Patent Owner”) timely filed a Preliminary Response (Paper 6, “Prelim.
`Resp.”). Pursuant to our authorization, Petitioner filed a Reply to Patent
`Owner’s Preliminary Response (Paper 7, “Petitioner’s Reply to Patent
`Owner’s Preliminary Response,” “Pet. Reply”) and Patent Owner filed a
`Sur-reply (Paper 9, “Patent Owner Sur-reply to Petitioner’s Reply to the
`Preliminary Response,” “PO Sur-reply”).
`We have authority to institute an inter partes review only if the
`information presented in the Petition shows “there is a reasonable likelihood
`that the petitioner would prevail with respect to at least 1 of the claims
`challenged in the petition.” 35 U.S.C. § 314(a) (2018). Upon consideration
`of the papers and evidence of record, we conclude that Petitioner has not
`shown a reasonable likelihood of prevailing with respect to at least one of
`the challenged claims. For the reasons explained below, we deny instituting
`an inter partes review on the challenged claims of the ’960 Patent.
`
`Related Matters
`B.
`The parties identify the following matters as related to this case:
`Uniloc USA, Inc. v. Amazon.com, Inc., no. 2-18-cv-00332 (E.D. Tex.);
`Uniloc 2017 LLC v. Google LLC, no. 2-18-cv-00551 (E.D. Tex.) (“Texas
`proceeding”). Pet. 2; Prelim. Resp. 4. The parties indicate that the Texas
`proceeding has been transferred to the Northern District of California per
`
`2
`
`
`
`IPR2020-00757
`Patent 7,012,960 B2
`court order dated June 19, 2020. Prelim. Resp. 4; Pet. Reply 1 (citing Ex.
`1019).
`
`The ’960 Patent
`C.
`The ʼ960 Patent, titled “Method of Transcoding and Transcoding
`Device with Embedded Filters,” is directed to “a method of transcoding a
`primary encoded signal comprising a sequence of pictures, into a secondary
`encoded signal.” Ex. 1001, 1:7–9. The transcoding method comprises a
`decoding step, which comprises a dequantizing sub-step for producing a first
`transformed signal. Id. at 1:11–13. Following the decoding step, the
`method includes an encoding step, which comprises a quantizing sub-step,
`for obtaining the secondary encoded signal. Id. at 1:14–16.
`By way of background, the ʼ960 Patent explains that bitrate
`transcoding allows a primary video stream to be encoded into a secondary
`video stream having a lower bitrate, “the bitrate reduction being performed
`in order to meet requirements imposed by the means of transport during
`broadcasting.” Id. at 1:24–29. The ʼ960 Patent describes that, in consumer
`devices, transcoding has often involved encoding “at a variable bitrate with a
`low average bit-rate” (id. at 1:51–56), and that prior art transcoding methods
`may lead to conspicuous quantization artifacts (id. at 1:59–60). To
`overcome this drawback, the ʼ960 Patent describes a transcoding method
`that “further comprises a filtering step between the dequantizing sub-step
`and the quantizing sub-step.” Id. at 1:64–67. The transcoding method in
`accordance with the invention allows filters to be implemented at negligible
`cost in the prior art transcoder. Id. at 2:1–3. These filters can be tuned to
`control the static and dynamic resolution and also to effect noise reduction.
`Id. at 2:3–5. For the same number of bits, the filtered transformed signal is
`
`3
`
`
`
`IPR2020-00757
`Patent 7,012,960 B2
`encoded with a smaller quantization scale, thus reducing visual artifacts such
`as blocking, ringing, and mosquito noise. Id. at 2:5–8.
`In a first embodiment of the invention of the ʼ960 Patent, the
`transcoder implements a motion-compensated temporal filter. Id. at 3:49–
`51. Figure 2 of the ’960 Patent is reproduced below:
`
`
`Figure 2 depicts a transcoder (200) comprising a decoding channel, an
`encoding channel, a prediction channel, and a temporal filter circuit Wt (21).
`Id. at 5:17–41. The decoding channel comprises a variable length decoder
`VLD (11) and a first dequantizer IQ (12) for decoding a current picture of a
`primary encoded signal (S1) and for producing a first transformed signal
`(R1). Id. at 5:18–21. The encoding channel comprises a quantizer Q (13), a
`variable length encoder VLC (14) for obtaining the secondary encoded
`signal (S2), and a second dequantizer IQ (15) for delivering a second
`transformed signal (R2). Id. at 5:22–25. The prediction channel comprises a
`subtractor (201), for determining a transformed encoding error (Re) and
`whose negative input receives the second transformed signal, an inverse
`discrete cosine transform circuit IDCT (16), a picture memory MEM (17), a
`
`4
`
`
`
`IPR2020-00757
`Patent 7,012,960 B2
`circuit for motion-compensation MC (18), a discrete cosine transform circuit
`DCT (19), for predicting a transformed motion-compensated signal (Rmc),
`and an adder (202), for delivering a sum of the transformed motion-
`compensated signal and the first transformed signal (R1) to the positive
`input of the subtractor. Id. at 5:26–37. Temporal filter circuit Wt (21)
`receives said sum and delivers the filtered transformed signal (Rt) to the
`quantizer Q (13). Id. at 5:38–40.
`In the second and third embodiments of the invention of the ʼ960
`Patent, the transcoder implements a spatial filter. Id. at 5:52–53. Figure 4 of
`the ’960 Patent is reproduced below:
`
`
`Figure 4 is a transcoder according to the third embodiment of the invention,
`with spatial post-filtering whose weight factors are Wsi j. Id. at 7:15–16.
`Transcoder (400) comprises a decoding channel, an encoding channel, a
`prediction channel, and a spatial filter circuit Ws (41). Id. at 7:17–40. The
`decoding channel comprises a variable length decoder VLD (11) and a first
`dequantizer IQ (12) for producing a first transformed signal (R1). Id. at
`7:18–20. The encoding channel, comprises a quantizer Q (13), a variable
`
`5
`
`
`
`IPR2020-00757
`Patent 7,012,960 B2
`length encoder VLC (14) and a second dequantizer IQ (15), and further
`comprising an inverse filter circuit (42) for producing a second transformed
`signal (R2). Id. at 7:21–25. The prediction channel comprises a subtractor
`(201), for determining a transformed encoding error (Re) and whose
`negative input receives the second transformed signal, an inverse discrete
`cosine transform circuit IDCT (16), a picture memory MEM (17), a circuit
`for motion-compensation MC (18), a discrete cosine transform circuit DCT
`(19), for predicting a transformed motion-compensated signal (Rmc), and an
`adder (202), for delivering a sum of said transformed motion-compensated
`signal and the first transformed signal (Rl) to the positive input of the
`subtractor. Id. at 7:26–37. Spatial filter circuit Ws (41) receives said sum
`and delivers a filtered transformed signal (Rf) to the encoding channel. Id.
`at 7:38–40.
`Finally, in a fourth embodiment of the invention of the ʼ960 Patent the
`transcoder further comprises a switch with at least two positions. Id. at
`7:43–7:65. Figure 5 depicting the fourth embodiment is reproduced below:
`
`
`
`6
`
`
`
`IPR2020-00757
`Patent 7,012,960 B2
`In a first position, a spatial filter Ws receives the output of the adder and
`delivers a filtered transformed signal to the quantizing circuit (13). Id. at
`7:64–8:1. Unlike the second and third embodiments, the spatial filter here is
`not applied to every macroblock in the current picture, but is only applied to
`intra-coded macroblocks in said picture. Id. at 8:1–5. In a second position
`of the switch, no filtering is applied and this position corresponds mainly to
`non-intra-coded macroblocks. Id. at 8:5–7. A third position of the switch
`corresponds to a temporal filter Wt that receives the output of the adder. Id.
`at 8:8–12. The temporal filter is applied to non-intra-coded macroblocks
`and spatial filtering is applied to intra-coded macroblocks. Id. at 8:12–15.
`
`Illustrative Claims
`D.
`Of the challenged claims, claims 1 and 4 are independent. Claim 5
`depends from independent claim 4.
`Claims 1 and 4 are reproduced below with annotations.
`1.
`[1.a] A method of transcoding a primary encoded signal
`(S1) comprising a sequence of pictures, into a secondary encoded
`signal (S2), said method of transcoding comprising at least the
`steps of:
`[1.b] decoding a current picture of the primary encoded signal,
`[1.c] said decoding step comprising a dequantizing sub-step (12)
`for producing a first transformed signal (R1),
`[1.d] encoding, following the decoding step, for obtaining the
`secondary encoded signal,
`[1.e] said encoding step comprising a quantizing sub-step (13),
`[1.f] wherein said method of transcoding further comprises a
`filtering step between the dequantizing sub-step and the
`quantizing sub-step, said filtering step using a recursive filter
`[1.g] wherein the recursive filtering step is intended to use a
`recursive filter such as: Rf[i]=(1—.alpha.[i]) (R1[i]+Rmc[i]),
`
`7
`
`
`
`IPR2020-00757
`Patent 7,012,960 B2
`where Rf[i], R1[i] and Rmc[i] are transformed coefficients
`comprised in the transformed signals (Rf,R1,Rmc) and .alpha.[i]
`is a filter coefficient comprised between 0 and 1; and
`[1.h] predicting a transformed motion-compensated signal from
`a transformed encoding error derived from the encoding step,
`[1.i] said prediction step being situated between the encoding and
`decoding steps,
`[1.j] wherein the recursive filtering step is for receiving the
`transformed motion-compensated
`signal
`and
`the
`first
`transformed signal and for delivering a filtered transformed
`signal to the quantizing sub-step.
`
`
`
`[4.a] A method of transcoding a primary encoded signal
`4.
`comprising a sequence of pictures, into a secondary encoded
`signal, said method of transcoding comprising at least the steps
`of:
`[4.b] decoding a current picture of the primary encoded signal,
`[4.c] said decoding step comprising a dequantizing sub-step for
`producing a first transformed signal,
`[4.d] encoding, following the decoding step, for obtaining the
`secondary encoded signal,
`[4.e] said encoding step comprising a quantizing sub-step,
`[4.f] wherein said method of transcoding further comprises a
`filtering step between the dequantizing sub-step and the
`quantizing sub-step; and
`[4.g] predicting a transformed motion-compensated signal from
`a transformed encoding error derived from the encoding step,
`[4.h] said prediction step being situated between the encoding
`and decoding steps,
`[4.i] wherein the filtering step is a spatial filtering step for
`receiving the transformed motion-compensated signal and the
`first transformed signal and for delivering a filtered transformed
`signal to the quantizing sub-step,
`
`8
`
`
`
`IPR2020-00757
`Patent 7,012,960 B2
`[4.j] said spatial filtering step being only applied to intra-coded
`macroblocks contained in the current picture.
`Ex. 1001, 9:27–53, 9:64–10:19.
`
`Prior Art and Asserted Challenges
`E.
`Petitioner asserts that claims 1, 4, and 5 would have been unpatentable
`based on the following challenges:
`Ground
`Claim(s)
`Challenged
`
`1
`2
`3
`4
`
`1
`1
`4, 5
`4, 5
`
`35 U.S.C. §
`103(a)
`103(a)
`103(a)
`103(a)
`
`Reference(s)/Basis
`Keesman1, Neri2
`Keesman, Neri, Dubois3
`Keesman, Kim,4
`Keesman, Kim,
`Matsumura5
`
`
`In addition, Petitioner relies on the Declaration of Jeffrey J. Rodriguez,
`Ph.D. (Ex. 1002) in support of the asserted challenges of unpatentability.
`
`
`
`1 Keesman et al., Transcoding of MPEG bitstreams, Signal Processing:
`Image Communication, Vol. 8, No. 6 (September 1996) (Ex. 1005,
`“Keesman”).
`2 Neri et al., Inter-block filtering and downsampling in DCT domain, Signal
`Processing: Image Communication, Vol. 6, No. 4 (August 1994) (Ex. 1006,
`“Neri”).
`3 Dubois & Sabri, Noise Reduction in Image Sequences Using Motion-
`Compensated Temporal Filtering, IEEE Transactions on Communications,
`Vol. Com-32, No. 7 (July 1984) (Ex. 1007, “Dubois”).
`4 Kim, US 6,249,549 B1, June 19, 2001 (Ex. 1008, “Kim”).
`5 Matsumura et al., US 6,792,045 B2, Sept. 14, 2004 (Ex. 1009,
`“Matsumura”).
`
`9
`
`
`
`IPR2020-00757
`Patent 7,012,960 B2
`
`II. ANALYSIS
`Level of Ordinary Skill in the Art
`A.
`Petitioner argues that a person of ordinary skill in the art at the time of
`the invention “would have had a Bachelor’s degree in Electrical
`Engineering, Computer Science, or the equivalent thereof, and two or more
`years of experience with data compression systems and algorithms,
`including video coding.” Pet. 7 (citing Ex. 1002 ¶¶ 16–21). Petitioner also
`asserts that “[m]ore education can supplement practical experience and vice
`versa.” Pet. 7. Patent Owner does not set forth a position on the level of
`ordinary skill in the art at this time, but reserves the right to propose its own
`definition if trial is instituted. Prelim. Resp. 12.
`For purposes of this Decision, we adopt Petitioner’s description,
`which we determine to be consistent with the asserted prior art references
`and the ’960 Patent. At this stage, Petitioner’s proposal is supported by
`record evidence (i.e., Dr. Rodriguez’s testimony) and Patent Owner did not
`submit evidence regarding the level of ordinary skill.
`
`Claim Construction
`B.
`In this case, we apply the claim construction standard set forth in
`Phillips v. AWH Corp., 415 F.3d 1303 (Fed. Cir. 2005). 37 C.F.R.
`§ 42.100(b); see also Changes to the Claim Construction Standard for
`Interpreting Claims in Trial Proceedings Before the Patent Trial and Appeal
`Board, 83 Fed. Reg. 51,340 (Oct. 11, 2018). Under the Phillips standard,
`claim terms must be given “the meaning that the term would have to a
`person of ordinary skill in the art in question at the time of the invention.”
`Phillips, 415 F.3d at 1313. Only claim terms in controversy require express
`construction, “and only to the extent necessary to resolve the controversy.”
`
`10
`
`
`
`IPR2020-00757
`Patent 7,012,960 B2
`Nidec Motor Corp. v. Zhongshan Broad Ocean Motor Co., 868 F.3d 1013,
`1017 (Fed. Cir. 2017).
`Petitioner proposes constructions for several claim limitations:
`“transformed signal[s]” (Claims 1 and 4) and “transformed coefficients”
`(Claim 1), “transformed motion-compensated signal” (Claims 1 and 4), and
`“wherein the recursive filtering step is intended to use a recursive filter such
`as: Rf[i]=(1— .alpha.[i]) (R1[i]+Rmc[i]) . . . is a filter coefficient comprised
`between 0 and 1.” Pet. 11–13. Patent Owner argues “the Petition is
`impermissibly keyed to incorrect claim constructions.” Prelim. Resp. 13.
`As demonstrated in the analysis below, for purposes of this Decision,
`we determine the terms proposed by Petitioner do not require express
`construction to determine whether Petitioner has met the “reasonable
`likelihood” threshold standard for institution of a trial.
`C. Grounds under § 103(a)
`Petitioner contends claim 1 is obvious in view of Keesman and Neri;
`claim 1 is obvious in view of Keesman, Neri, and Dubois; claims 4 and 5 are
`obvious in view of Keesman and Kim; and claims 4 and 5 are obvious in
`view of Keesman, Kim, and Matsumura. Pet. 3–4.
`A claim is unpatentable under § 103(a) if the differences between the
`claimed subject matter and the prior art are “such that the subject matter as a
`whole would have been obvious at the time the invention was made to a
`person having ordinary skill in the art to which said subject matter pertains.”
`KSR Int’l Co. v. Teleflex Inc., 550 U.S. 398, 406 (2007). The question of
`obviousness is resolved on the basis of underlying factual determinations,
`including: (1) the scope and content of the prior art; (2) any differences
`between the claimed subject matter and the prior art; (3) the level of skill in
`the art; and (4) when in evidence, objective evidence of nonobviousness, i.e.,
`
`11
`
`
`
`IPR2020-00757
`Patent 7,012,960 B2
`secondary considerations. Graham v. John Deere Co., 383 U.S. 1, 17–18
`(1966).
`Additionally, the obviousness inquiry typically requires an analysis of
`“whether there was an apparent reason to combine the known elements in
`the fashion claimed by the patent at issue.” KSR, 550 U.S. at 418 (citing
`In re Kahn, 441 F.3d 977, 988 (Fed. Cir. 2006) (requiring “articulated
`reasoning with some rational underpinning to support the legal conclusion of
`obviousness”)); see In re Warsaw Orthopedic, Inc., 832 F.3d 1327, 1333
`(Fed. Cir. 2016) (citing DyStar Textilfarben GmbH & Co. Deutschland KG
`v. C.H. Patrick Co., 464 F.3d 1356, 1360 (Fed. Cir. 2006)).
`As explained below, we determine based on the present record that
`Petitioner has not shown a reasonable likelihood it would prevail in
`establishing that each of claims 1, 4, and 5 are unpatentable as obvious under
`35 U.S.C. § 103(a).
`
`D.
`
`Alleged Obviousness over Keesman and Neri
`Overview of Keesman
`1.
`Keesman relates to the transcoding problem of bit-rate conversion,
`with a concentration on transcoding Moving Picture Experts Group (MPEG)
`signals into MPEG signals. Ex. 1005, 481–482.6 Keesman discusses the
`problems of reducing the complexity of a transcoder and picture quality loss.
`Id. at 482. Keesman notes that a typical source of performance loss in a
`system including transcoding is the double occurrence of a quantization
`operation. Id. Figure 1 of Keesman, reproduced below, depicts the basic
`configuration of a system including a transcoder:
`
`
`6 All page numbers refer to the original internal page numbers of the
`Exhibits.
`
`12
`
`
`
`IPR2020-00757
`Patent 7,012,960 B2
`
`
`Figure 1 depicts an encoder that compresses the incoming video signal at a
`bit-rate of R1. Id. at 482. The compressed signal is converted into a
`compressed format of a lower bit-rate R2 and finally a decoder decompresses
`the incoming signal and displays the resulting video signal. Id.
`Figure 7 of Keesman is reproduced below:
`
`
`Figure 7 depicts a transcoder that includes a decoder, an extra DCT, an extra
`quantizer, de-quantizer, and a VLC with a buffer. Id. at 486–88. Keesman
`teaches that transmitting a video signal across a system as shown in Figure 1
`
`13
`
`
`
`IPR2020-00757
`Patent 7,012,960 B2
`may lead to extra distortion compared to directly compressing the video
`signal at the final bit-rate R2. Id.
`The transmission chain model of Figure 1 is shown in greater detail in
`Figure 9, reproduced below:
`
`
`
`The pictures are put through the DCT, quantized, dequantized, quantized,
`dequantized, and finally put through an IDCT. Id. at 488.
`Overview of Neri
`2.
`Neri relates to image coding and specifically to the use of DCT
`techniques. Ex. 1006, 303 (Abstract). Neri describes filtering and
`downsampling methods directly acting on the DCT domain. Id. Neri
`explains that “DCT image transforms usually operate on blocks” and, thus,
`“it is useful that the DCT filtering techniques preserve the block dimension.”
`Id. Neri teaches that “[f]iltering and downsampling of image signals are
`basic operations normally performed in image processing.” Id. at 303. Neri
`discloses that filtering and downsampling are usually carried out in the space
`domain. Id. However, Neri describes using filtering and downsampling
`directly in the transformed (DCT) domain. Id.
`Analysis of Claim 1
`3.
`Petitioner contends claim 1 is obvious over Keesman and Neri under
`35 U.S.C. § 103. Pet. 3, 14–55. Petitioner argues Keesman and Neri teach
`each of the limitations of claim 1, including limitation 1.f which recites
`
`14
`
`
`
`IPR2020-00757
`Patent 7,012,960 B2
`“wherein said method of transcoding further comprises a filtering step
`between the dequantizing sub-step and the quantizing sub-step, said filtering
`step using a recursive filter.” Pet. 30–42. For this limitation, Petitioner
`acknowledges that “Keesman does not disclose a filtering step using a
`recursive filter between the dequantizing sub-step and the quantizing sub-
`step (between the dequantizer DQ1 and quantizer Q2),” but argues that it
`would have been obvious to combine the teachings of Keesman and Neri
`such that a recursive filter is placed between the dequantizing and quantizing
`sub-steps of Keesman. Pet. 30 (citing Ex. 1002 ¶ 80).
`According to Petitioner, Neri teaches a DCT-based inter-block filter
`that is used to reduce artifacts, aliasing effects, and distortion when
`transcoding video data. Pet. 32 (citing Ex. 1006, Abstract, 304, 316).
`Petitioner argues the output of Keesman’s dequantizing step (DQ1) is
`dequantized transform coefficients. Pet. 34 (citing Ex. 1002 ¶ 85). Because
`Neri’s DCT-based inter-block filter operates on DCT coefficients, Petitioner
`argues Neri’s filter would operate on Keesman’s signal. Pet. 34 (citing Ex.
`1005, Fig. 7, 9; Ex. 1002 ¶ 85).
`Petitioner argues that one of ordinary skill in the art would have
`positioned Neri’s filter after Keesman’s adder circuit and before Keesman’s
`quantizer Q2 as depicted in Figure 7 of Keesman, because such a
`configuration would be most applicable to filtering DCT coefficients and it
`would provide transformed coefficients that correspond to compensated
`block data. Pet. 34 (citing Ex. 1002 ¶ 86, Ex. 1005, Figs. 7, 9). This
`position would be “between the dequantizing sub-step and the quantizing
`sub-step” of Keesman. Pet. 35.
`Specifically, Petitioner argues there are “about four locations in
`Keesman’s transcoding chain . . . where the signal corresponds to
`
`15
`
`
`
`IPR2020-00757
`Patent 7,012,960 B2
`coefficients in the frequency domain (e.g., DCT) which Neri’s DCT filter
`could operate on.” Pet. 36 (citing Ex. 1005, Fig. 7). The four locations
`referred to by Petitioner can be seen in an annotated version of Keesman’s
`Figure 7, depicted in the Petition. See Pet. 37. We reproduce that figure
`here for ease of reference.
`
`
`The above figure illustrates four locations where Petitioner argues Neri’s
`filter could be placed in Keesman’s transcoding system where the signal
`corresponds to coefficients in the frequency domain.
` Petitioner contends that, because “there were a limited number of
`options . . . where a filter such as Neri’s DCT-based inter-block filter could
`be placed in Keesman’s system, all of which provide the predictable result of
`filtering compensated block data in the frequency domain,” a person of
`ordinary skill “would have had reason to try to place Neri’s DCT filter after
`the adder circuit following the dequantizing sub-step.” Pet. 36 (citing
`Ex. 1002 ¶¶ 87–88). Petitioner’s expert testifies that, in the case of P or B
`macroblocks, two of the four positions after the adder circuit (positions 3
`and 4 in the above figure) would be preferable because the DCT data would
`
`16
`
`
`
`IPR2020-00757
`Patent 7,012,960 B2
`be incomplete at the two locations before the adder circuit (i.e. at positions 1
`and 2). See Ex. 1002 ¶¶ 88–89. Petitioner argues that “it would have been
`nothing more than common sense to include a filter step between the
`decoding step and encoding step to reduce distortion introduced during the
`transcoding.” Pet. 41.
`According to Petitioner, the placement of Neri’s filter after Keesman’s
`dequantizing sub-step would make the filter a recursive filter because the
`output of the filter depends on the previous output of the filter via a feedback
`loop. Pet. 38 (citing Ex. 1002 ¶ 90). Petitioner argues that one of ordinary
`skill in the art would have been motivated to implement Neri’s filter as a
`recursive filter to improve its processing efficiency. Pet. 40 (citing Ex. 1002
`¶ 92). Petitioner argues that it is well known that implementing a moving
`average filter, such as Neri’s DCT filter, as a recursive filter would improve
`its processing speed by using data from the input and previously calculated
`data from the output. Pet. 40 (Ex. 1002 ¶ 92).
`Patent Owner argues that neither Keesman nor Neri disclose or
`suggest the use of a recursive filter at the specific position recited in claim 1.
`Prelim. Resp. 25. Patent Owner argues that there is no evidence that Neri’s
`filter could not be implemented at a different position than the four options
`identified by Petitioner, or that it could not be implemented in another
`system altogether. Prelim. Resp. 26. Because of this, Patent Owner argues
`that Petitioner relies on impermissible hindsight reconstruction in arguing
`that the claimed invention is obvious. Prelim. Resp. 25–26.
`On the present record, we are unpersuaded Petitioner has shown a
`reasonable likelihood that the combination of Keesman and Neri teach
`limitation 1.f. Specifically, we find Petitioner has not presented sufficient
`evidence that would support a finding that Neri’s filter would be
`
`17
`
`
`
`IPR2020-00757
`Patent 7,012,960 B2
`implemented as a recursive filter and be placed between the dequantizing
`sub-step and the quantizing sub-step of Keesman’s transcoding system. As
`Patent Owner argues, Keesman and Neri do not explicitly teach that a filter
`be used in a transcoding system at the claimed location. Nor does Petitioner
`argue that either reference suggests the use of a filter at that location.
`Furthermore, neither reference teaches or suggests using a recursive filter.
`To fill these gaps, Petitioner relies on expert testimony in arguing that
`there are “about” four locations where the signal corresponds to coefficients
`in the frequency domain which Neri’s DCT filter could operate on, “all of
`which provide the predictable result of filtering compensated block data in
`the frequency domain.” Pet. 36. Yet, Petitioner argues that it would be
`obvious to one of ordinary skill in the art to place Neri’s filter in one of
`those locations, specifically location 3, rather than the others. See Pet. 35
`(showing a figure with the filter placed after Keesman’s adder circuit but
`before quantizer Q2). Since the motivation provided by Petitioner for adding
`Neri’s filter to Keesman is to reduce distortion, and Petitioner states that all
`four locations would provide that benefit, more explanation is needed to
`narrow the choices available to the claimed location (location 3).
`Petitioner’s expert testifies that, in the case of predictive macroblocks
`(P and B macroblocks), one of ordinary skill would have chosen locations 3
`and 4 over locations 1 and 2. Ex. 1002 ¶ 88. Location 4 does not satisfy the
`claim limitations as it is after the quantization sub-step of Keesman and,
`therefore, not “between the dequantizing sub-step and the quantizing sub-
`step” as claimed. Petitioner’s expert, however, does not provide a reason to
`further narrow the choice to location 3 over location 4. Thus, even after
`relying on its expert’s testimony, Petitioner has not fully established that one
`of ordinary skill would place Neri’s filter at the claimed location over the
`
`18
`
`
`
`IPR2020-00757
`Patent 7,012,960 B2
`other locations that Petitioner argues Neri’s filter could equally have been
`placed.
`Petitioner argues that it would be a matter of design choice and
`common sense to try to place Neri’s filter at the claimed location, but
`common sense is better invoked when providing a motivation to combine
`references, rather than supplying a missing claim limitation. Arendi S.A.R.L.
`v. Apple Inc., 832 F.3d 1355, 1361 (Fed. Cir. 2016). Here the location of the
`filter is an explicit limitation of the claim, and we find that Petitioner has not
`provided sufficient reasoning and analysis in conjunction with reliance on
`common sense to demonstrate that Neri’s filter would be placed in exactly
`the claimed location in Keesman’s system.
`Accordingly, on the present record, Petitioner has not established a
`reasonable likelihood it would prevail in showing that claim 1 is
`unpatentable over the combination of Keesman and Neri.
`
`E.
`
`Alleged Obviousness over Keesman, Neri, and Dubois
`Overview of Dubois
`1.
`Dubois is titled “Noise Reduction in Image Sequences Using Motion-
`Compensated Temporal Filtering” and is directed to a nonlinear temporal
`filtering algorithm using motion compensation for reducing noise in image
`sequences. Ex. 1007, 826. Dubois describes a nonlinear recursive filtering
`approach extended by the application of motion compensation techniques.
`Id. Figure 1 of Dubois is reproduced below:
`
`19
`
`
`
`IPR2020-00757
`Patent 7,012,960 B2
`
`
`Figure 1 depicts a block diagram of a first-order recursive temporal filter
`with motion compensation. Id. at 827. This filter requires a frame memory
`in order to be able to form the prediction, and a module for estimating the
`displacement field. Id. The displacement estimator can use the input signal
`as well as any of the signals available in the noise reducer to perform the
`estimate. Id.
`
`Analysis of Claim 1
`2.
`Under the alleged obviousness ground relying on Keesman and Neri,
`Petitioner argued that limitation 1.g is entitled to no patentable weight. Pet.
`42–44. Petitioner argues, in the alternative, that if limitation 1.g is given
`patentable weight, then Dubois, in combination with Keesman and Neri,
`teaches limitation 1.g. Pet. 55–60. In doing so, Petitioner relies on the same
`arguments and evidence presented above in the obviousness ground relying
`on Keesman and Neri for all of the limitations of claim 1, except limitation
`1.g. Importantly, Petitioner does not rely on Dubois to argue that the prior
`art references teach limitation 1.f which requires a recursive filter to be
`positioned “between the dequantizing sub-step and the quantizing sub-step.”
`Instead Petitioner relies on the same arguments and evidence for limitation
`1.f.
`
`Accordingly, for the same reasons explained above in our analysis of
`the obviousness ground relying on Keesman and Neri, we find Petitioner has
`
`20
`
`
`
`IPR2020-00757
`Patent 7,012,960 B2
`not established a reasonable likelihood that Keesman, Neri, and Dubois
`teach the limitations of claim 1.
`
`F.
`
`Alleged Obviousness over Keesman and Kim
`Overview of Kim
`1.
`Kim is titled “Down Conversion System Using a Pre-Decimation
`Filter” and is directed to “a decoder which converts and formats an encoded
`high resolution video signal, e.g. MPEG-2 encoded video signals, to a
`decoded lower resolution output video signal, and more specifically to a
`down conversion system for the decoder.” Ex. 1008, code (54), 1:5–9.
`Kim’s Figure 2 is reproduced below.
`
`
`Kim’s Figure 2 “is a high level block diagram of an exemplary embodiment
`of a down conversion system.” Id. at 4:55–56.
`Analysis of Claim 4
`2.
`Limitation 4.f
`a)
`Limitation 4.f recites “wherein said method of transcoding further
`comprises a filtering step between the dequantizing sub-step and the
`quantizing sub-step.” Ex. 1001, 4:6–9. This limitation is similar to
`
`21
`
`
`
`IPR2020-00757
`Patent 7,012,960 B2
`limitation 1.f, which required a recursive filter to be placed “between the
`dequantizing sub-step and the quantizing sub-step.” Id. at 9:38–39.
`Petitioner presents similar arguments here as it does for limitation 1.f.
`For example, Petitioner acknowledges that Keesman does not disclose a
`filtering step, but argues that one of ordinary skill in the art would have
`found it obvious to combine Keesman with Kim such that a filter would be
`placed between the dequantizing and quantizing sub-steps. Pet. 62.
`Petitioner argues Kim “discloses an ‘intra-block lowpass filter[]’ that is
`applied to the coefficients in the frequency domain.” Pet. 62 (citing
`Ex. 1008, 5:11–12) (alteration in original). According to Petitioner, an
`inverse quantizer 214 provides DCT coefficients to Kim’s DCT filter 216.
`Pet. 63 (citing Ex. 1008, 5:48–52, Fig. 2). Thus, Petitioner argues, Kim’s
`DCT filter 216 would equally apply to dequantized transformed coefficients
`in Keesman. Pet. 64 (citing Ex. 1002 ¶ 134).
`Petitioner argues that one of ordinary skill in the art would have
`positioned Kim’s filter after Keesman’s adder circuit following the decoding
`unit because such a configuration would be most ap