UNITED STATES PATENT AND TRADEMARK OFFICE BEFORE THE PATENT TRIAL AND APPEAL BOARD GOOGLE INC., Petitioner v. VEDANTI SYSTEMS LIMITED, Patent Owner Case IPR2016-00212 Patent 7,974,339 B2 PETITIONER S OPPOSITION TO PATENT OWNER S MOTION TO AMEND Mail Stop PATENT BOARD Patent Trial and Appeal Board U.S. Patent & Trademark Office P.O. Box 1450 Alexandria, VA 22313-1450
TABLE OF CONTENTS TABLE OF AUTHORITIES... ii I. Introduction... 1 II. Argument... 2 A. Vedanti s Motion fails to provide a proper construction for new claim terms having language not found in the specification.... 3 B. Vedanti s Motion fails to demonstrate patentability of the proposed substitute claims with respect to 35 U.S.C. 112.... 5 1. The specification of the 339 patent fails to provide adequate written description and enablement for the new claim elements.... 5 2. The terms of the proposed substitute claims are indefinite... 8 a. Uniform matrix size data... 8 b. Non-predetermined... 9 C. Vedanti s Motion fails to show patentability under 35 U.S.C. 103 of the proposed substitute claims over the known material prior art.... 11 1. Vedanti s motion fails to demonstrate patentability of the claim language itself.... 11 2. Vedanti s motion fails to demonstrate patentability over the prior art for generating optimized matrix data from the frame data, wherein the optimized matrix data defines at least two regions having different aspect ratios. (Claims 14-15)... 13 3. Vedanti s motion fails to demonstrate patentability over the prior art for selecting a non-predetermined set of pixel data. (Claims 16-17)... 18 D. Vedanti s motion fails to demonstrate patentability under 35 U.S.C. 103 of substitute claims 14 and 15 over other material prior art... 20 i. Spriggs in view of Golin, and further in view of Shin... 21 ii. Shin in view of Spriggs... 23 III. Conclusion... 25 - i -
TABLE OF AUTHORITIES Supreme Court Cases KSR Int l Co. v. Teleflex Inc., 550 U.S. 398 (2007)... 16, 17 Federal Circuit Cases ClassCo, Inc. v. Apple, Inc., 838 F.3d 1214 (Fed. Cir. 2016)... 16, 17 In re Gordon, 733 F.2d 900 (Fed. Cir. 1984)... 16 In re Ratti, 270 F.2d 810 (CCPA 1959)... 16 Inphi Corp. v. Netlist, Inc., 805 F.3d 1350 (Fed. Cir. 2015)...7, 8 LizardTech v. Earth Res. Mapping Inc., 424 F.3d 1336 (Fed. Cir. 2005)... 7 Microsoft Corp. v. Proxyconn, Inc., 789 F.3d 1292 (Fed. Cir. 2015)... 3 Oakley, Inc. v. Sunglass Hut Int l, 316 F.3d 1331 (Fed. Cir. 2003)... 5 Santarus, Inc. v. Par Pharma., Inc., 694 F.3d 1344 (Fed. Cir. 2012)... 7 Board Decisions CaptionCall, L.L.C. v. Ultratec, Inc., IPR2015-00637, Paper 98 (PTAB 2016).. 7, 8, 9 Ex parte Miyazaki, 89 USPQ2d 1207 (BPAI 2008)... 9, 10 Idle Free Sys., Inc. v. Bergstrom, Inc., IPR2012-00027, Paper 26 (PTAB 2014)...2, 3 JDS Uniphase Corp. v. Fiber, LLC, IPR2013-00318, Paper 45 (PTAB 2014)... 5 MasterImage 3D, Inc. v. RealD Inc., IPR2015-00040, Paper 42 (PTAB 2015)... 13 Statutes 35 U.S.C. 103... 5, 11, 20 35 U.S.C. 112...3, 5 35 U.S.C. 112(a)... 5, 6, 8 35 U.S.C. 112(b)... 5, 10 Regulations 37 C.F.R. 42.20(c)... 5 - ii -
Petitioner Google Inc. ( Google ) hereby opposes Patent Owner s Contingent Motion to Amend (Paper 16 ( Mot. )) by Vedanti Systems Limited ( Vedanti ). I. Introduction Vedanti filed a Motion to Amend that seeks to substitute claims 14-17 in place of canceled claims 7, 9, 10, and 12, respectively. Claim 14 adds a requirement for different aspect ratios a term without support in the 339 specification. Claim 16 adds that regions are defined by region data, and Vedanti contrives an indefinite term non-predetermined set of pixel data in its attempt to avoid prior art. As the evidence here shows, none of these limitations adds anything patentable to the independent claims. Claims 15 and 17 also add nothing patentable, as they simply correspond to issued claims 9 and 12 respectively, with just their dependencies updated. The Motion to Amend is itself deficient in several respects and should be denied. The Motion does not provide a proper claim construction for two new claim limitations: different aspect ratios and non-predetermined. This is troublesome, as these terms are not in the original specification, lack adequate written description support and enablement, and are indefinite. Vedanti also fails to demonstrate the patentability of the proposed substitute claims over known material prior art, including Spriggs, Golin, and Keith. This failure is acute, as Vedanti overlooks teachings of blocks with different aspect - 1 -
ratios in Spriggs (GOOG 1005), and a specific teaching in Golin (GOOG 1006) (and Keith (Ex. 2015)) describing randomly selecting pixels. Vedanti alleges a lack of motivation to combine Spriggs and Golin (or Keith), but falls short. Vedanti s arguments are misdirected at fill codes taught in Golin and Keith, rather than the actual block division teachings used to render the claims obvious. In this way, Vedanti has not shown claims 14-17 to be patentable over Spriggs in view of Golin (or Keith). Further, substitute claims 14-15 would have been obvious based on new material prior art by Shin. Shin (GOOG 1035) teaches subdividing an image into blocks with different aspect ratios, and is properly combinable with Spriggs in view of Golin to render claims 14-15 obvious. Alternatively, claims 14-15 are unpatentable using Shin as a primary reference combined with the pixel selection teaching of Spriggs. Because Vedanti fails to set forth a prima facie case for the relief requested or satisfy its burden of proof, its motion should be denied in its entirety. II. Argument In a motion to amend, the Patent Owner bears the burden of proof to demonstrate patentability of its proposed substitute claims over the prior art, and, thus, entitlement to add these claims to its patent. Idle Free Sys., Inc. v. Bergstrom, Inc., IPR2012-00027, Paper 26, at 7 (PTAB 2014); see also Microsoft Corp. v. - 2 -
Proxyconn, Inc., 789 F.3d 1292, 1304-06 (Fed. Cir. 2015). As the Board stated in Idle Free: The burden is not on the petitioner to show unpatentability, but on the patent owner to show patentable distinction over the prior art of record and also prior art known to the patent owner. Some representation should be made about the specific technical disclosure of the closest prior art known to the patent owner, and not just a conclusory remark that no prior art known to the patent owner renders obvious the proposed substitute claims. Idle Free, Paper 26, at 7. Vedanti has not met its burden. The evidence shows that the proposed substitute claims are unpatentable on multiple grounds. A. Vedanti s Motion fails to provide a proper construction for new claim terms having language not found in the specification. The flaws in Vedanti s Motion begin with its introduction of two new claim terms different aspect ratios and non-predetermined that are not found in the patent s specification. The Motion s flaws are compounded by Vedanti s failure to provide claim constructions for these terms, or explain what the proper claim constructions are under a BRI. The Board has held that a motion to amend claims must identify how the proposed substitute claims are to be construed, especially when the proposed substitute claims introduce new claim terms. JDS Uniphase Corp. v. Fiber, LLC, IPR2013-00318, Paper 45, at 47 (PTAB 2014) - 3 -
(citing Idle Free, Paper 26, at 7). In particular, Vedanti adds the terms different aspect ratios (Claim 14) and non-predetermined (Claim 16), even though neither term appears in the 339 Patent specification or underlying PCT application. Vedanti admits as much in its Motion on pages 3-4, where it provides no citations or references for either term. At most, the Motion alleges support based on limited examples in the specification, despite the claims using different language. (Id.) For example, for different aspect ratios, as recited in claim 14, Vedanti points to Fig. 10 and its accompanying text. (See Mot. 3.) Vedanti also points to general text on non-uniform sizes and nonsymmetrical shapes (id.), but none of the text explicitly recites the term aspect ratio, much less the term different aspect ratios. Nor does Vedanti provide a proper claim construction for the new term. Vedanti instead muddies the water by characterizing claim 14 as requiring non-uniform aspect ratios (Mot. 2, 13), which is different from the actual language of claim 14. In a similar vein, for non-predetermined (Claim 16), Vedanti points to text stating that pixels can be selected (1) in accordance with a ʻpredetermined sequence or location, (2) randomly, or (3) using some other suitable selection criteria. (Id. at 4.) From this series, Vedanti appears to assume that the term nonpredetermined is supported by reference to a random selection. Not only is this incorrect, but Vedanti still does not provide a proper construction under a BRI for - 4 -
the term non-predetermined, much less identify whether it is limited or not limited to random selection. Like the patent owner in JDS, Vedanti failed to provide any explicit construction of new claim terms introduced by its Motion. See JDS Uniphase Corp. IPR2013-00318, Paper 45, at 47-48 (PTAB 2014). Claim construction is an important step in patentability determination. Id. at 46 (citing Oakley, Inc. v. Sunglass Hut Int l, 316 F.3d 1331, 1339 (Fed. Cir. 2003)). Because Vedanti has failed to take this crucial first step, Vedanti has failed to meet its burden to demonstrate patentability of substitute claims 14-17 under 37 C.F.R. 42.20(c). For these reasons alone, Vedanti s Motion should be denied because a proper claim construction was not provided. In the absence of a claim construction, Vedanti cannot show patentability of its substitute claims 14-17 under at least 35 U.S.C. 112(a), 112 (b) and 103. Even apart from its lack of claim construction, Vedanti still has not met its burden to show that substitute claims 14-17 are patentable under at least 35 U.S.C. 112(a), 112 (b) and 103. Each of these failures is described in turn below. B. Vedanti s Motion fails to demonstrate patentability of the proposed substitute claims with respect to 35 U.S.C. 112. 1. The specification of the 339 patent fails to provide adequate written description and enablement for the new claim elements. Vedanti has not shown that claims 14-17 are adequately described and - 5 -
enabled under 35 U.S.C. 112(a). For claim 14, Vedanti relies entirely on the specification at col. 10, lines 38-47, and Fig. 10 as allegedly supporting at least two regions having different aspect ratios. (Mot. 2-3 (citing GOOG 1018 67).) However, nowhere in the drawings, the specification, or the original claims does the term aspect ratio appear, let alone the term different aspect ratios. Instead, written description merely states, [e]ach matrix of the array [of pixel data] can be of different size, but the matrices must form the array when combined. (GOOG 1001, 10:39-41.) Different size does not require different aspect ratios. For example, the specification states that the size of matrices within the frame can be varied, such that a given frame is made up of matrices varying in size, such as from a 1 1 matrix to a 5 5 matrix or greater. (Id., 4:1-3; GOOG 1030 94.) Here, different sizes maintain the same aspect ratio of 1:1. Similarly, the later described non-symmetrical matrices of different sizes still have the same, uniform aspect ratio N:M. (GOOG 1001, 4:3-6.) The 339 patent has only a single example in Fig. 10 where selected matrices may be considered to have proportionally different numbers of horizontal and vertical pixels. This sole example does not isolate two different aspect ratios as such, and does not convey to a POSA that the inventor possessed a claimed invention covering all possible frames with at least two regions having different aspect ratios. (Cf. GOOG 1030 93.) Thus, the 339 patent does not clearly convey that the inventor had - 6 -
possession of the claimed subject matter at the time of filing and would enable one of ordinary skill to practice the full scope of substitute claims 14 and 15. LizardTech v. Earth Res. Mapping Inc., 424 F.3d 1336, 1346 (Fed. Cir. 2005) (citations omitted). Accordingly, the motion to amend should be denied at least with respect to claims 14 and 15. Substitute claim 16 includes the new term non-predetermined. This negative limitation does not appear anywhere in the disclosure of the 339 patent. For this newly added negative limitation to have proper written description support, the disclosure must at least list reasons for selecting a non-predetermined set, or specify alternatives where applicable. See CaptionCall, L.L.C. v. Ultratec, Inc., IPR2015-00637, Paper 98, at 108-09 (PTAB 2016) (citing Inphi Corp. v. Netlist, Inc., 805 F.3d 1350 (Fed. Cir. 2015) and Santarus, Inc. v. Par Pharma., Inc., 694 F.3d 1344 (Fed. Cir. 2012). Here, the specification only discloses predetermined sequence, predetermined order, and predetermined location with respect to pixel selection in a predefined matrix. (GOOG 1001, 4:11-30.) The negative limitation of selecting a non-predetermined set of pixel data, or any supporting rationale, is nowhere to be found in the specification or by any reasonable inference from a POSA. As such, the disclosure is unclear as to why random selection or other suitable manners (id.) would be an alternative to the predetermined means specified. Introducing the non-predetermined language - 7 -
appears to be an arbitrary dissection of the 339 patent amending the claims in order to avoid the prior art, which the Board and the Federal Circuit have disallowed. CaptionCall, IPR2015-00637, Paper 98, at 109 (quoting Inphi, 805 F.3d at 1356). Lacking support, the new language of substitute claims 16 and 17 constitutes impermissible new matter. 2. The terms of the proposed substitute claims are indefinite. a. Uniform matrix size data Considering substitute claim 15 as a whole with the limitations of claim 14, the term uniform matrix size data is indefinite. According to Vedanti, claim 14 s recitation, wherein the optimized matrix data defines at least two regions having different aspect ratios, refers directly to Fig. 10 of the patent. (E.g., GOOG 1001, 10:38-47; see also Mot. 3; GOOG 1001, 5:54-66.) This stands in contrast to the uniform matrix size of Fig. 9, recited in claim 15. (E.g., GOOG 1001, 10:19-23 ( Each matrix of the array is of uniform size, such as 4 4. Thus the matrix size data for the entire frame can be represented by a single data set. ).) Because regions having different aspect ratios are non-uniform (see Mot. 2, 13), claim 15 requires a frame to have both uniform and non-uniform matrix size data at the same time. (See GOOG 1030 97.) Yet this is not possible. (Id.) Because of this conflicting claim scope, substitute claim 15 is further indefinite and unpatentable. - 8 -
b. Non-predetermined The term non-predetermined in substitute claims 16 and 17 is indefinite, as not clearly reciting the metes and bounds of the claimed invention and subject to multiple meanings. A POSA could not determine the scope of nonpredetermined. This is not surprising, because the specification gives no guidance on the term s meaning. (See id.) The non-predetermined term was simply contrived by Vedanti to avoid prior art (see CaptionCall, IPR2015-00637, Paper 98, at 109). The scope of non-predetermined could refer to not known ahead of time based on its ordinary meaning, or it could refer to random, given the only other example in the 339 specification that is not predetermined. The Board has held that claims are indefinite where an element is amenable to two plausible definitions. Ex parte Miyazaki, 89 USPQ2d 1207, 1215 (BPAI 2008) (precedential). Such ambiguity may exist when a definition in the specification differs from the ordinary meaning of the words in the claim, and where neither the specification, nor the claims, nor the ordinary meanings of the words provides any guidance as to what [Patentee] intends to cover with this claim language. Id. The first plausible definition is the ordinary meaning of nonpredetermined, which signifies something not already determined before a selection event occurs. Vedanti s own expert supports this construction, swearing - 9 -
that this term means not known ahead of time. (GOOG 1034 at 125:8-14.) On the other hand, Vedanti also argues that the term is supported by the specification s implication of what the term is not. The specification includes random and other suitable manners as other possibilities compared to predetermined. (See Mot. 4 (citing GOOG 1018, 25, 37-40, 46, 57-60).) 1 This gives rise to a different yet equally plausible interpretation that nonpredetermined means random. This ambiguity is further compounded by not explaining how the term other suitable manners affects the meaning of nonpredetermined. Indeed, language similar to other suitable means is used in the 339 patent to describe predetermined examples that are outside the scope of the substitute claims (GOOG 1001, 4:17-18 and 4:41-43). Under Miyazaki, because substitute claims 16-17 recite or incorporate a limitation amenable to multiple ambiguous definitions that cannot be resolved, they are indefinite under 35 U.S.C. 112(b). Vedanti s Motion to Amend should thus be denied as to claims 16 and 17. 1 Because the 339 patent also fails to define what is random, the ordinary meaning in the art must govern, which is that of lacking order or predictability. (See GOOG 1033 at 373 (1999).) This definition is not necessarily a basis to form a clear understanding of the term non-predetermined. - 10 -
C. Vedanti s Motion fails to show patentability under 35 U.S.C. 103 of the proposed substitute claims over the known material prior art. With respect to prior art, Vedanti has not met its burden to show patentability over at least two material references (Golin, GOOG 1006, and Keith, Ex. 2015). Both of these references were well-known to Vedanti, as they were either applied in this proceeding or cited in the original patent prosecution. Vedanti fails to direct its patentability analysis to the claim language itself, and ignores material teachings having text that is identical to the newly added claim language. Even if these deficiencies are ignored, Vedanti fails to substantively show that its substitute claims are patentable. In addressing obviousness of the new claims over Spriggs and Golin (and other art), Google builds on the assumption that Spriggs and Golin, as cited in this proceeding, have already rendered the corresponding original claim elements obvious, and that the new limitations in Vedanti s Motion to Amend are newly considered here in combination with the original claim elements. 1. Vedanti s motion fails to demonstrate patentability of the claim language itself. As mentioned above, substitute claims 14 and 15 now recite at least two regions having different aspect ratios. Vedanti admits in its Motion that Golin and Keith at Fig. 27 teach a subregion 2706 that has a different aspect ratio than the other subregions. (Mot. 7.) Vedanti later mischaracterizes Golin and Keith as - 11 -
teaching non-uniform aspect ratios (id.) rather than different aspect ratios. By doing this, Vedanti admits that different aspect ratios are taught by each of Golin and Keith, but does not explain why such different aspect ratios are not combinable. Vedanti only provides analysis (albeit incorrect) for why nonuniform aspect ratios cannot be combined with Spriggs. Failing to address material teachings in Golin and Keith for claims 16 and 17 is even more telling. Substitute claims 16 and 17 recite selecting a nonpredetermined set of pixel data from each region, and Vedanti explicitly points to random pixel selection as one example that provides support for the nonpredetermined claim term. (Mot. 4.) Yet Vedanti does not discuss relevant parts of Golin and Keith that read directly on this new language. Instead, Vedanti generically argues that both Golin and Keith do[] not teach or suggest pixel selection. (Mot. 22.) Such an argument is incorrect. Golin and Keith both explicitly disclose a pixel randomly selected from each of these sub-regions. (GOOG 1006, 28:1-2; Ex. 2015, 27:63-64 (emphasis added).) Vedanti offers no explanation of non-obviousness to account for these parts of Golin and Keith, which clearly read on Vedanti s substitute claims 16 and 17. The Motion to Amend should thus be denied in its entirety for filing to address the material prior art of record. Vedanti questions the publication dates of Exhibits 2006-2008, but fails to - 12 -
discuss their teachings on the merits. Further, Vedanti has not submitted or made any specific reference to the material prior art of record from the prosecution of the patent, e.g., U.S. Pat. Nos. 5,778,092, 5,784,175, 5,838,333, 5,878,169, 6,014,181, 6,078,307, 6,198,467, 6,326,981, 6,473,062, 6,608,632, and 7,551,189, which are cited on the face of the 339 patent, along with European patent documents and other publications. Vedanti glosses over these additional references, stating only that Patent Owner does not believe that any disclose the claim language. (See Mot. 11.) Vedanti provides no further discussion of the references teachings, even in general terms. As such, Vedanti has failed to meet its duty of candor and good faith by grouping prior art references together according to their particular teachings. MasterImage 3D, Inc. v. RealD Inc., IPR2015-00040, Paper 42, at 3 (PTAB 2015). Even if these flaws in Vedanti s analysis are ignored, the evidence still does not show non-obviousness of the proposed substitute claims over Spriggs in view of Golin (or Keith). 2. Vedanti s motion fails to demonstrate patentability over the prior art for generating optimized matrix data from the frame data, wherein the optimized matrix data defines at least two regions having different aspect ratios. (Claims 14-15) As established on the record of this proceeding, Spriggs in view of Golin renders obvious claims 7, 9, 10 and 12. The newly proposed substitute claims - 13 -
Patent No. 7,974,3399 B2 would similarly have been obvious over Spriggs in view of Golin. Indeed, for claims 14-15, Vedanti admits that the new limitations are taught by Golin and only challenges the combination of the teachings. Spriggs discloses that transmitted data comprisess a sequence of division codes, each indicating whether or not the corresponding area is divided; each division code being followed by the codes corresponding to parts of the respective area. (GOOG 1005, Fig. 6, 5:27-31 (shown in part immediately below).) Under a BRI, Spriggs vertices define att least two regions having different aspect ratios. Spriggs, Fig. 6. For example, the rectangles defined by FBIG versus JLCH or EPJR, each having a different aspect ratio, would have been obvious. In this example, JLCH is four times as wide as EPJR and half the height off FBIG, and EPJR is one fourth as wide as FBIG and JLCH and half the height of FBIG. (See, e.g., GOOG 1030 95; GOOGG 1034, 102:6-15, 125:15-126:4.) Spriggs even teaches sending sample values with each division code includingg 0s and 1s.. (GOOG 1005, 3:2-12.) Even if Spriggs alone may not have rendered substitute claim 14 obvious to a POSA, Golin in combination with Spriggs would have, for at least the following reasons: Fig. 27 of Golin shows binary tree regionalization of an image, which - 14 -
Patent No. 7,974,3399 B2 divides a region into sub-regions, either horizontally or vertically, depending on edge detection in parts of the region. A first split in Fig. 27 is horizontal, followed by a second split below the horizontal split, vertically dividing the bottom half. Binary tree regionalizationn is the preferred mode because it has been found to normally result in fewer regions and hence fewerr bits. (GOOG 1006, 13:46-48.) As shown in Fig. 27 (right), for example, the right side of split 2 (2706) is a rectangular region having an aspect ratio that is half the width off split 1 s top side of (2704), and twice the height off split 3 s bottom side (2710). Because the heights and widths of the other rectangles are not proportional to those of 2706, these rectangles are the result of a binary tree regionalization generating optimized matrix data from the frame data (e.g., id.,, Figs. 50-51), which defines at least two regions having different aspect ratios, as claimed in substitute claim 14. A POSA possessing the teachings of Spriggs and Golin at the time of alleged invention would have seen Golin s binary tree divisions as a simple substitution for Spriggs division codes, with a predictable result. (GOOGG 1030 105-109.) For example, being able to use regions of different aspect ratios would have improved the accuracy of coding regions as taught by Golin to be able to differentiate - 15 -
additional splits as horizontal or vertical, depending on the number of edges in a region to be subdivided. (GOOG 1030 106.) Simply by accounting for the different dimensions of multiple subregions, as one of ordinary skill in the art would have done, this modification would have yielded further reductions in data transmitted. This would have improved transmission data size, compared to simply keeping splits having the same aspect ratio. This would not have rendered Golin inoperable for its intended purpose, let alone change its principle of operation. Cf. In re Gordon, 733 F.2d 900, 902 (Fed. Cir. 1984); In re Ratti, 270 F.2d 810, 813 (CCPA 1959). Vedanti admits that Golin and Keith contain the new teachings of claim 14, but attacks the motivation to combine with Spriggs. Vedanti contends that [o]ne of ordinary skill in the art would have had no incentive to apply the non-uniform aspect ratios disclosed by Golin and Keith in Spriggs, because when the objective is to reduce data, non-uniform aspect ratio would increase the necessary division code data. (Mot.,6-7.) This is overly restrictive, as it views the references in a piecemeal fashion and fails to consider the proper motivation a POSA would have to combine the teachings of Golin or Keith with Spriggs. As the Supreme Court noted in KSR, [a] person of ordinary skill is also a person of ordinary creativity, not an automaton. KSR Int l Co. v. Teleflex Inc., 550 U.S. 398, 421 (2007). KSR does not require that a combination only unite old - 16 -
elements without changing their respective functions. ClassCo, Inc. v. Apple, Inc., 838 F.3d 1214, 1219 (Fed. Cir. 2016). KSR instructs that the obviousness inquiry requires a flexible approach. Id. (citing KSR, 550 U.S. at 415). Under ClassCo, even if different aspect ratios would increase division code data, there are other considerations that would have made this combination obvious to try. See ClassCo, 838 F.3d at 1219. Vedanti improperly assumes that Spriggs was rigidly designed with a simple split/no split flag, and thus unable to support different aspect ratios. (Mot. 7.) Vedanti maintains that a POSA would have lacked motivation to use different aspect ratios, because it would have added bits to the data transmission. (Id.) Yet, Vedanti s own expert, Dr. Kia, admitted that the number of bits needed to transmit matrix data defining different aspect ratios would be small on the order of six bits or less. (GOOG 1034, 42:1 47:12.) Vedanti employs the same type of rigid argument that the Federal Circuit warned against in ClassCo: KSR explains that the ordinary artisan recognizes that familiar items may have obvious uses beyond their primary purposes, and in many cases a person of ordinary skill will be able to fit the teachings of multiple patents together like pieces of a puzzle. [KSR, 550 U.S.] at 420. The rationale of KSR does not support ClassCo s theory that a person of ordinary skill can only perform combinations of a puzzle element A with a perfectly fitting puzzle element B. ClassCo, 838 F.3d at 1219. Far from Vedanti showing unpatentability, the - 17 -
rationales set out above show that a POSA would have been motivated to combine the teachings of Spriggs and Golin or Keith. Claim 15 recites the same feature as claim 9 for which it is substituted, only changing the dependency to depend from claim 14. Claim 15 thus adds no separately patentable features as proven with respect to claim 9. 3. Vedanti s motion fails to demonstrate patentability over the prior art for selecting a non-predetermined set of pixel data. (Claims 16-17) Vedanti argues for the patentability of substitute independent claim 16 by relying on the non-predetermined term to distinguish over Spriggs. Vedanti maintains that Spriggs uses a predetermined pixel selection, not a nonpredetermined selection. Vedanti points to one example where it claims that Spriggs transmits the same five center pixel values for the subregions. These pixels are always predetermined. (Mot. 20.) Vedanti s argument is based on a superficial (and erroneous) reading of Spriggs that ignores how the block subdivision in Spriggs is carried out. The entirety of Spriggs teaching makes clear that, for one embodiment, only the corners defining the whole frame or initial region may be known in advance. (GOOG 1005, 2:67 3:2.) Spriggs determines whether to divide subregions based on whether or not each further subdivided region can be interpolated from its corner values when encountering edges or fine detail. (Id., 2:55-60, 3:2-24.) Because this operation divides and subdivides each - 18 -
frame and region, it is not possible for the pixels to be always predetermined. The division process must be repeated for each frame, with different results for different images. Thus, selected pixels for regions cannot be known in advance, and therefore must be non-predetermined, as Vedanti s expert Dr. Kia would agree. (See GOOG 1034 at 125:8-14.) Accordingly, Spriggs teaches selecting a non-predetermined set of pixel data from each region to produce selection pixel data for each region. Vedanti s arguments with respect to Golin and Keith also fall short because they are misdirected at fill codes rather than the actual teachings relied on. (Mot. 22 ( Golin then determines, for each sub-region, a fill code for representing all of the pixel values of each sub-region. (Id. at 13:12-19.) However, Golin does not teach or suggest pixel selection. Keith (Ex. 2015) evinces the same deficiencies as Golin. ).) Fill codes may be relevant to other parts of Vedanti s claims, but what Vedanti cites is not relevant to substitute claim 16. Though overlooked by Vedanti and not addressed in its Motion, Golin does disclose a pixel randomly selected from each of these sub-regions. (GOOG 1006, 28:1-2.) Unlike the Spriggs teaching of selecting the same center pixel values for each subregion, Golin selects pixels for each subregion at random. (Id., 27:57 28:19.) Golin also provides motivation for one of ordinary skill in the art at the time to use random selection instead of predetermined selection, by explaining that - 19 -
[a] strategy of picking [] pixels at random, but uniformly distributed over the region [from each subregion] has proven to be effective in order to determine a representative set of pixels for [a mean square difference] calculation. (Id.) Thus, even if non-predetermined means random, the Spriggs-Golin combination would have suggested to a POSA random pixel selection from each region. Accordingly, substitute claim 16 is unpatentable over the prior art, and Vedanti has not met its burden to show patentability over the prior art of record. Claim 17 recites the same feature as claim 10 for which it is substituted, only changing the dependency to depend from claim 16. Claim 17 thus adds no separately patentable features as proven with respect to claim 10. D. Vedanti s motion fails to demonstrate patentability under 35 U.S.C. 103 of substitute claims 14 and 15 over other material prior art. It is not surprising that substitute claims 14 and 15 are unpatentable, since they add limitations drawn to different aspect ratios, which were well-known. The state of the art at the time of the 339 patent reveals many other patent references that use different sizes, shapes, and aspect ratios of regions of frames for image compression or data reduction for storage and transmission. Shin (GOOG 1035) teaches dividing blocks into different aspect ratios. Other notable examples include Jacobs et al. (U.S. Pat. 5,862,262, Fig. 3), Thorell (U.S. Pat. 6,373,988, Fig. 1e), Mehrotra (U.S. Pat. 6,571,016, Fig. 12), Ohtani (U.S. Pat. 6,671,321, - 20 -
Figs. 3-8), Tsougarakis (U.S. Pat. 6,901,110, 15:49-53), and Sawhney (U.S. Pat. 6,907,073, 5:44-53). (GOOG 1036-1041.) Accordingly, even if the combination of Spriggs and Golin were not sufficient to render the substitute claims obvious, which Google does not admit, substitute claims 14-15 would have been obvious over Spriggs in view of Golin, and further in view of Shin (GOOG 1035). Alternatively, substitute claims 14-15 would have been obvious over Shin in view of Spriggs. i. Spriggs in view of Golin, and further in view of Shin As established on the record of this proceeding, Spriggs in view of Golin renders obvious claims 7 and 9. The additional limitations in substitute claims 14 and 15 add nothing patentable, and would have been obvious over Spriggs and Golin further in view of Shin. Even if, arguendo, it would not have been immediately apparent to combine Golin s splits resulting in different aspect ratios (GOOG 1006, Fig. 27) with Spriggs division codes (0, 1), Shin expressly teaches making such a combination. Accordingly, it would have been obvious to combine the quad-tree regionalization of Shin to the Spriggs-Golin combination, to arrive at optimized matrix data that defines at least two regions having different aspect ratios, as recited in claim 14. Like Spriggs, Shin also uses division codes. Spriggs has a one-bit division code for each region/subregion, indicating whether or not there is a further division - 21 -
Patent No. 7,974,3399 B2 (GOOGG 1005, 3:2-38; Fig. 6). Shin uses a two-bit division code, indicating This extra bit in the division code results multiple regions having different aspect ratios. in the frame being subdivided (See, e.g., into GOOG 1035, Fig. 1, 5B, 7, 8B (showing rectangularr divisions, some of which are short and wide,, some off which are tall and narrow, and some of which are approximately square).) Shin simply adds one bit to the division code to save more data from transmissio on. (See id.., 8:45-52. ) Shin s transmission output equates to substitute claim 14 s optimized Given this explicit suggestion in Shin to achieve a whether or not to split the region, and if so, whether the split is horizontal, vertical, or both. (GOOG 1035, 7:32-43, 8:2-322 ( quadtree code value ).) ma- trix dataa from the frame dataa [that] defines at least two regions having different as- map and representative [pixel] value[s]. (Id., 3:54-59, 5: 35-40, 6:11-20, 8:38-44; see also id., 8:2-32, 8:45-52, Fig. 4 (showing transmission pect ratios, in the form of a bit stream array for outputting the generated quadtree output).) good quality image hav- - 22 -
ing no distortion, and expressed by only a small quantity of data as compared with the conventional method shown in Fig. 8A (id., 8:45-52), a POSA would easily have substituted Shin s two-bit division code for Spriggs one-bit division code. (GOOG 1030 110-113.) This would accommodate different aspect ratios with no unexpected results. (Id.) Shin contemplates subdivisions of a frame such as those taught by Spriggs. (GOOG 1035, Fig. 8A.) Shin also teaches the modification (id., Fig. 8B) that renders substitute claim 14 obvious. Claim 15 adds nothing patentable, for all the reasons given for claim 7 for which it is substituted. Moreover, the uniform matrix size data of substitute claim 15 can be found in the prior art teachings. (See, e.g., GOOG 1035, Fig. 5B 01 ; GOOG 1006, Figs. 32A-J; GOOG 1005, Fig. 2.) Accordingly, substitute claims 14 and 15 would have been obvious for at least these reasons over Spriggs and Golin and further in view of Shin. ii. Shin in view of Spriggs Substitute claims 14 and 15 are alternatively obvious over Shin in view of Spriggs. The combination of Shin and Spriggs teaches all limitations of the substitute claims. For substitute claim 14, Shin teaches a method for transmitting data, comprising: receiving frame data (GOOG 1035, 2:34 ( inputting image data of frame unit ), 3:61-64 ( frame memory for receiving frame image data )); generating optimized matrix data from the frame data (id., 3:54-59 ( quadtree map - 23 -
drawing means... to generate... a bit stream array for outputting the generated quadtree map and the representative value as a bit stream, whereby the quadtree map and representative values are optimized matrix data); see Figs. 6-7), wherein the optimized matrix data defines at least two regions having different aspect ratios. (Id., Figs. 1, 5A-7, 8B.) Shin also teaches that receiving frame data comprises receiving an array of pixel data (GOOG 1035, 2:47-50 ( image type of each block is a distribution value with respect to a plurality of pixel values within a block, and the representative value is an average value of pixels in each block ).) Shin also teaches selecting an average value of the pixels for each region. (GOOG 1035, 6:11-52; 8:29-44.) Vedanti may argue that Shin does not explicitly disclose selecting pixel data as recited, but Spriggs teaches selecting one of two or more sets of pixel data based on the optimized matrix data (GOOG 1005, 2:67-3:68 (pixel data that are selected for each block according to the corner coordinates, division codes, and corner addresses of the blocks (i.e., optimized matrix data)).) The references themselves suggest their combination. Spriggs teaches selecting pixel data in regions to reduce the amount of pixel data needing to be transmitted. Accordingly, it would have been obvious to a POSA to combine Spriggs pixel selection with Shin s optimized matrix data defining regions having different aspect ratios, rendering claim 14 obvious. (GOOG 1030 118-121.) - 24 -
Shin in combination with Spriggs also teaches that generating the optimized matrix data from the frame data comprises: setting a matrix size based on pixel selection data (GOOG 1035, Fig. 7 block size and Fig. 5B splits being of different sizes as defined by the 339 patent, GOOG 1001, 5:54-66, 10:38-40); and transmitting the selection pixel data and the optimized matrix data by assembling the optimized matrix data and the selection pixel data into a generated display frame. (GOOG 1035, Fig. 4: bit stream array is an output that represents a generated display frame for transmission, e.g., Fig. 7; GOOG 1005, 2:36-47.) This combination of Shin and Spriggs would have been a simple substitution yielding predictable results. (GOOG 1030 123-124.) The uniform matrix size data of substitute claim 15 can also be found in the prior art teachings. (See, e.g., GOOG 1035, Fig. 5B 01 ; GOOG 1005, Fig. 2.) For at least these reasons, substitute claims 14 and 15 would have been obvious over Shin in view of Spriggs, with all their relevant teachings and the level of ordinary skill in the art at the time the invention was made. III. Conclusion Because Vedanti fails to set forth a prima facie case for the relief requested or satisfy its burden of proof, its motion to amend should be denied in its entirety. - 25 -
Respectfully submitted, STERNE, KESSLER, GOLDSTEIN & FOX P.L.L.C. Date: November 21, 2016 /Michael Messinger/ Michael V. Messinger, Reg. No. 37,575 Michelle K. Holoubek, Reg. No. 54,179 Attorneys for Petitioner 1100 New York Avenue, N.W. Washington, D.C. 20005-3934 (202) 371-2600 - 26 -
CERTIFICATION OF SERVICE The undersigned hereby certifies that the foregoing PETITIONER S OPPOSITION TO PATENT OWNER S MOTION TO AMEND was served electronically via e-mail on November 21, 2016, in its entirety on the following: Robert M. Asher John J. Stickevers SUNSTEIN KANN MURPHY & TIMBERS LLP 125 Summer Street Boston, MA 02110 Tel: (617) 443-9292 Fax: (617) 443-0004 rasher@sunsteinlaw.com jstickevers@sunsteinlaw.com sunsteinip@sunsteinlaw.com Date: November 21, 2016 1100 New York Avenue, N.W. Washington, D.C. 20005-3934 (202) 371-2600 4886264_6.docx STERNE, KESSLER, GOLDSTEIN & FOX P.L.L.C. / Michael Messinger / Michael V. Messinger Attorney for Petitioner Registration No. 37,575 Michelle K. Holoubek Attorney for Petitioner Registration No. 54,179