Hyung Joon. KimDownload PDFPatent Trials and Appeals BoardAug 9, 201914083423 - (D) (P.T.A.B. Aug. 9, 2019) Copy Citation UNITED STATES PATENT AND TRADEMARK OFFICE UNITED STATES DEPARTMENT OF COMMERCE United States Patent and Trademark Office Address: COMMISSIONER FOR PATENTS P.O. Box 1450 Alexandria, Virginia 22313-1450 www.uspto.gov APPLICATION NO. FILING DATE FIRST NAMED INVENTOR ATTORNEY DOCKET NO. CONFIRMATION NO. 14/083,423 11/18/2013 Hyung Joon Kim TI-72976 8105 23494 7590 08/09/2019 TEXAS INSTRUMENTS INCORPORATED P O BOX 655474, M/S 3999 DALLAS, TX 75265 EXAMINER NOH, JAE NAM ART UNIT PAPER NUMBER 2481 NOTIFICATION DATE DELIVERY MODE 08/09/2019 ELECTRONIC Please find below and/or attached an Office communication concerning this application or proceeding. The time period for reply, if any, is set in the attached communication. Notice of the Office communication was sent electronically on above-indicated "Notification Date" to the following e-mail address(es): uspto@ti.com PTOL-90A (Rev. 04/07) UNITED STATES PATENT AND TRADEMARK OFFICE ____________ BEFORE THE PATENT TRIAL AND APPEAL BOARD ____________ Ex parte HYUNG JOON KIM ________________ Appeal 2018-0072681 Application 14/083,423 Technology Center 2400 _________________ Before JEAN R. HOMERE, JAMES B. ARPIN, and DAVID J. CUTITTA II, Administrative Patent Judges. ARPIN, Administrative Patent Judge. DECISION ON APPEAL Appellant2 appeals under 35 U.S.C. § 134(a) the Examiner’s final rejection of claims 1–3 and 11, all of the pending claims. App. Br. 2. Claims 4–10 and 12–20 are cancelled. Id. at 15–16 (Claims App’x). We have jurisdiction under 35 U.S.C. § 6(b). We affirm. 1 In this Decision, we refer to Appellant’s Appeal Brief (“App. Br.,” filed August 1, 2017) and Reply Brief (“Reply Br.,” filed April 16, 2018); the Final Office Action (“Final Act.,” mailed December 30, 2016); the Examiner’s Answer (“Ans.,” mailed February 16, 2018); and the originally filed Specification (“Spec.,” filed November 18, 2013). Rather than repeat the Examiner’s findings and determinations and Appellant’s contentions in their entirety, we refer to these documents. 2 Appellant asserts Texas Instruments Inc. is the real party-in-interest. App. Br. 2. Appeal 2018-007268 Application 14/083,423 2 STATEMENT OF THE CASE Appellant’s recited methods and computer readable media comprising instructions for performing such methods “generally relate [to] video coding and more specifically relate to adaptive coding unit (CU) partitioning based on image statistics.” Spec. ¶ 2. In particular, in [High Efficiency Video Coding (HEVC)], a largest coding unit (LCU) is the base unit used for block-based coding. A picture is divided into non-overlapping LCUs. That is, an LCU plays a similar role in coding as the macroblock of H.264/AVC, but it may be larger, e.g., 32x32, 64x64, etc. An LCU may be partitioned into coding units (CU) using recursive quadtree partitioning. A CU is a block of pixels within an LCU and the CUs within an LCU may be of different sizes. The quadtree is split according to various criteria until a leaf is reached, which is referred to as the coding node or coding unit. The maximum hierarchical depth of the quadtree is determined by the size of the smallest CU (SCU) permitted. Id. ¶ 20; see id. ¶ 19 (listing various versions of HEVC); see also Sjoberg ¶ 48 (splitting the LCU in “quadtree fashion”). As noted above, claims 1–3 and 11 are pending. Claims 1 and 11 are independent. App. Br. 15–16 (Claims App’x). Claims 2 and 3 depend directly or indirectly from claim 1. Id. As noted above, claim 11 recites computer-readable media “comprising instructions that, when executed by at least one processor, cause the at least one processor to” perform methods of computing at least one statistical measure and selecting one of the LCU or CUs to form an LCU partitioning based on the statistical measure, including the methods recited in claim 1. Id. at 16. Therefore, we focus our analysis on the overlapping limitations of these independent claims. See Accenture Global Servs. GmbH v. Guidewire Software, Inc., 728 F.3d 1336, 1341 (Fed. Cir. 2013) (“Although CLS Bank issued as a plurality opinion, in that case a Appeal 2018-007268 Application 14/083,423 3 majority of the court held that system claims that closely track method claims and are grounded by the same meaningful limitations will generally rise and fall together.” (citation omitted)). Claim 1, reproduced below with disputed limitation emphasized, is representative. 1. A method for determining coding unit (CU) partitioning of a largest coding unit (LCU) of a picture, the method comprising: computing, with one or more processors, a first statistical measure and a second statistical measure for the LCU based on pixel values of the LCU and without using residual pixel values of the LCU; selecting, with the one or more processors, the LCU as the CU partitioning when the first statistical measure does not exceed a first threshold and the second statistical measure does not exceed a second threshold; and selecting, with the one or more processors, CUs in one or more lower layers of a CU hierarchy of the LCU to form the CU partitioning when the first statistical measure exceeds the first threshold and/or the second statistical measure exceeds the second threshold. App. Br. 15 (Claims App’x) (emphasis added). REFERENCES AND REJECTIONS The Examiner relies upon the following references: Name3 Number Publ’d Filed Porter US 2005/0128306 A1 Jun. 16, 2005 Dec. 10, 2004 Imamura US 2008/0253454 A1 Oct. 16, 2008 Apr. 11, 2008 Sjoberg US 2013/0003868 A1 Jan. 3, 2013 Nov. 9, 2011 Hebel US 2014/0086314 A1 Mar. 27, 2014 Sep. 26, 2012 3 All reference citations are to the first named inventor only. Appeal 2018-007268 Application 14/083,423 4 The Examiner rejects claims 1–3 and 11 under 35 U.S.C. § 103 as rendered obvious over the combined teachings of Sjoberg and Hebel. Final Act. 5–9. The Examiner also rejected claims 6–8, 12, 13, 17–20 as rendered obvious over the combined teachings of Sjoberg and Hebel; claims 4, 9, and 14 as rendered obvious over the combined teachings of Sjoberg, Hebel, and Porter, and claims 5, 10, and 15 as rendered obvious over the combined teachings of Sjoberg, Hebel, Porter, and Imamura. Id. at 8–11. Further, the Examiner asserted claims 6–10 included means-plus-function limitations per 35 U.S.C. § 112(f). Id. at 4–5. Because Appellant cancelled claims 4–10, 12–15, and 17–204 in an Amendment filed with the Appeal Brief, these additional rejections and the means-plus-function interpretations applied to claims 6–10 now are moot. App. Br. 2. We review the appealed rejections for error based upon the issues identified by Appellant, and in light of the arguments and evidence produced thereon. Ex parte Frye, 94 USPQ2d 1072, 1075 (BPAI 2010) (precedential). Arguments not made are waived. See 37 C.F.R. § 41.37(c)(1)(iv). Unless otherwise indicated, we adopt the Examiner’s findings in the Final Action and the Answer as our own and add any additional findings of fact for emphasis. We address the rejection below. 4 Although the Examiner refers to claim 16 in the Final Office Action, Appellant cancelled claim 16 in an Amendment filed on November 18, 2016. Final Act. 9. Appeal 2018-007268 Application 14/083,423 5 ANALYSIS Obviousness Over Sjoberg and Hebel 1. Independent Claims 1 and 11 Claims 1 and 11 stands rejected as unpatentable over the combined teachings of Sjoberg and Hebel. Final Act. 5–8, 9. The Examiner determines that Sjoberg teaches the steps of computing, with one or more processors, a first statistical measure . . . for the LCU based on pixel values of the LCU and without using residual pixel values of the LCU; selecting, with the one or more processors, the LCU as the CU partitioning when the first statistical measure does not exceed a first threshold . . . and selecting, with the one or more processors, CUs in one or more lower layers of a CU hierarchy of the LCU to form the CU partitioning when the first statistical measure exceeds the first threshold and/or the second statistical measure exceeds the second threshold,5 as recited in claim 1. Id. at 6–7 (emphasis added) (citing Sjoberg ¶¶ 5, 51); see id. at 9 (discussing claim 11). In particular, with regard to the selection of CU sizes Sjoberg teaches, This decision whether to split a coding unit is based on the coding process. For instance, a picture area that represents a fairly homogenous background is more efficiently represented using large CU sizes, such as LCUs, as compared to splitting the 5 Because the final limitation in claim 1 describes using a comparison of the first statistical measure to the first threshold “and/or” a comparison of the second statistical measure to the second threshold, we understand that only the first comparison is necessary to satisfy this limitation. App. Br. 15 (Claims App’x); see, e.g., Ex parte Gross, Appl’n No. 11/565,411, Appeal No. 2011-004811, at 4 (PTAB Jan. 3, 2014) (“We agree with Appellant that ‘and/or’ covers embodiments having element A alone, element B alone, or elements A and B taken together.”) Appeal 2018-007268 Application 14/083,423 6 picture area into smaller coding units. However, picture areas with small details or a lot of such details can generally not be correctly represented if using large coding units. In such a case, it is more efficient and preferred from coding quality point of view to use several smaller CUs for the picture area. The selection of whether to further split a CU can thereby [be] performed according to techniques described in the art and preferably based on the coding efficiency and quality. Sjoberg ¶ 51(emphasis added); see id. ¶ 48 (describing quadtree hierarchical splitting). In view of Sjoberg’s teachings, the Examiner asserts splitting the LCU depending upon the homogeneous or detailed nature of the pixels of the LCU (i.e.[,] which is understood to use values of pixels and not necessarily using residual pixel values). The mention of the coding efficiency and quality in Sjoberg is interpreted to be in the context of such homogeneous characteristic[s] of the LCU. That is for example, it would be more efficient to use LCU as the CU when the LCU is a homogeneous background, as disclosed by Sjoberg. Final Act. 2–3 (emphasis added). The Examiner acknowledges that Sjoberg does not teach, “computing . . . a second statistical measure for the LCU” and “selecting the LCU . . . when the second statistical measure does not exceed a second threshold.” Id. at 7. Nevertheless, the Examiner determines that Hebel teaches these limitations. Id. (citing Hebel ¶¶ 30, 44). In particular, the Specification explains, “[i]n some embodiments, the two statistical measures are the variance of the CU and the gradient of the CU.” Spec. ¶ 23; see id., claim 4 (Claim 4 (now cancelled) depends from claim 1 and recites “the first statistical measure is variance and the second statistical measure is gradient.”). Hebel teaches “[c]haracteristics of a coding unit considered in this determination may include, but are not limited to, texture, brightness, motion, variance, or any combination thereof.” Hebel ¶ 44 (emphases Appeal 2018-007268 Application 14/083,423 7 added). Further, a person of ordinary skill in the art would have understood the concepts of variance and gradient to be closely related. See Spec. ¶¶ 50 (determination of “variance”); 51 (determination of “gradient”); 64 (“One of ordinary skill in the art will understand embodiments in which an alternative statistical measurement is used. For example, the previously described variance computation includes multiplication, which will require more logic gates in a hardware implementation. Instead of computing variance, the simple sum of absolute differences between pixel values and the average pixel value may be computed. This latter statistic computation can be implemented with fewer logic gates than the full variance computation.” (emphasis added)); Porter ¶ 333 (“Note that the variance measure used can either be traditional variance or the sum of differences of neighbouring pixels (gradient) or any other variance-type measure.” (emphases added)); see also Final Act. 3 (“The use of the particular measure such as gradient is disclosed by the combined teaching of Sjoberg-Porter or Sjoberg-Hebel- Porter.”). Although the Examiner only relied upon Porter in the rejection of now-cancelled claim 4, claim 4 remains part of the Specification, and Porter remains part of the record of this appeal. We determine that claim 4’s disclosure and Porter’s teachings are relevant to a person of ordinary skill’s understanding of the meaning of gradient and variance and, consequently, to the breadth of the term “statistical measure.” See In re Morris, 127 F.3d 1048, 1054 (Fed. Cir. 1997) (“In construing this limitation, we apply the broadest reasonable meaning of the words in their ordinary usage, as those words would be understood by one of ordinary skill in the art, taking into account any definitions supplied by [Appellant’s] Specification.”); Okajima Appeal 2018-007268 Application 14/083,423 8 v. Bourdeau, 261 F.3d 1350, 1355 (Fed. Cir. 2001) (“[T]he absence of specific findings on the level of skill in the art does not give rise to reversible error ‘where the prior art itself reflects an appropriate level and a need for testimony is not shown’”) (quoting Litton Indus. Prods., Inc. v. Solid State Sys. Corp., 755 F.2d 158, 163 (Fed. Cir. 1985)). The Examiner concludes that a person of ordinary skill in the art would have had reason to combine the teachings of Sjoberg, regarding “computing” of a first statistical measure based on the homogeneity or detail of an image and “selecting” a CU based on that measure, with the teachings of Hebel, regarding the “computing” a second statistical measure based on characteristics, such as “variance,” to achieve the methods recited in claim 1. Final Act. 7–8. In particular, the Examiner concludes a person of ordinary skill in the art would be motivated [to combine the teachings of these references] as considering multiple measures to characterize an image area, as taught by Hebel, would more accurately reflect whether the image area is homogeneous or with small details when selecting the size of the coding units, which would result in the overall increase in the coding efficiency. Id. Appellant contends that the Examiner erred in the rejection of independent claims 1 and 11 as rendered obvious over the combined teachings of Sjoberg and Hebel for at least three reasons. App. Br. 5–11. For the reasons given below, we are not persuaded that the Examiner erred. First, Appellant contends that Sjoberg’s disclosure of determining CUs based on whether the image is homogeneous or detailed (Sjoberg ¶ 51) Appeal 2018-007268 Application 14/083,423 9 does not teach “computing” a statistical measure (see App. Br. 5–7; Reply Br. 3–4). In particular, Appellant submits that the Examiner erred because paragraph [0005] of Sjoberg et al. is merely giving a general observations about how different CU sizes may be used to handle both large and small homogenous areas. Similarly, paragraph [0051] of Sjoberg et al. is merely directed to a general observation about which types of pictures may benefit from using large CU sizes vs. small CU sizes. There is no indication that the homogeneity or detailed-naturedness of pictures is used in Sjoberg et al. to split an LCU. App. Br. 7. The Examiner asserts that The teaching of the disclosure of Sjoberg should be ascertained or understood in view of the level of the ordinary skilled in the art. Thus, given that Sjoberg discloses ([¶ 51]) use of large coding units for homogeneous picture and smaller coding units for detailed pictures for efficiency in coding, Sjoberg teaches that the splitting of the coding units into smaller ones should be based upon whether the picture is homogeneous or detailed. In fact, Sjoberg [¶ 51] expressly discloses splitting the coding unit to realize a smaller coding unit. Ans. 15–16; see Final Act. 2–3. Splitting the LCU into smaller CUs may be accomplished by the quadtree technique taught by Sjoberg (Sjoberg ¶ 48; Figs. 3A, 3B) and disclosed in the Specification (Spec., Figs. 2, 6). As the Specification and cancelled claim 4 make clear, variance and gradient may be statistical measures. Spec. ¶ 23, 50–53, 55, 64–66, claim 4. As Hebel and Porter teach CUs may be determined based on variance (Hebel ¶ 44), and variance and gradient, i.e., “the sum of differences of neighbouring pixels” (Porter ¶ 333), are both “variance-type measure[s]” (id.). Moreover, a person of ordinary skill in the art would understand a homogeneous image is one in which the differences between neighbouring Appeal 2018-007268 Application 14/083,423 10 pixels are necessarily small, e.g., lower variance or a shallow gradient, but an image having detail or small detail necessarily exhibits larger differences between neighbouring pixels, e.g., higher variance or a steep gradient. See Porter ¶¶ 325–332 (describing comparison of skin color variance to a threshold value), 376–384 (describing Figs. 29a–29c and “a gradient-based measure [] derived from the window . . . which is the average of the absolute differences between all adjacent pixels 1011 in both the horizontal and vertical directions, taken over the window”); see also RANDOM HOUSE WEBSTER’S COLLEGE DICTIONARY, 629 (2nd Random House ed. 1999) (“homogeneous” defined as “composed of parts or elements that are all of the same kind; not heterogeneous” and “of the same kind or nature; essentially alike”). Thus, we are persuaded that a person of ordinary skill in the art, having knowledge evidenced by Porter’s teachings, would understand that determining a CU based on image area homogeneity or detail, as taught by Sjoberg, in light of known CU characteristics, including variance, as taught by Hebel, encompasses the recited “statistical measure[s].” Second, Appellant contends “Hebel . . . is not concerned with LCU partitioning, and therefore fails to overcome the deficiencies already discussed above with respect to Sjoberg . . . .” App. Br. 6, 9. Nevertheless, the Examiner relies on Hebel to teach the computation of a second statistical measure and the selection of a CU based on the comparison of the second statistical measure to a second threshold. Final Act. 7. As noted above, Hebel teaches the use of known CU characteristics, “includ[ing], but are not limited to, texture, brightness, motion, variance, or any combination thereof,” to select a CU. Hebel ¶ 44 (emphases added). Thus, the Examiner Appeal 2018-007268 Application 14/083,423 11 does not rely on Hebel to supply what is taught by Sjoberg. Final Act. 6–7. Appellant improperly challenges the references individually, rather than the combination of the references’ teachings, upon which the Examiner relies. Third, Appellant contends that “[t]o the extent that efficiency and quality are statistical measures to categorize an image as having a homogenous background (which Appellant does not concede), such measures as already discussed above, are determined using residual pixels values of a CU,” which is contrary to the recitation of claims 1 and 11 that the CU is determined “without using residual pixel values of the LCU.” App. Br. 7; see Reply Br. 2–3. The Examiner asserts, “Appellant does not state specific reason why the appellant believes that Sjoberg . . . is ‘using residual pixels values of a CU.’” Ans. 15. In response, Applicant notes that Sjoberg . . . describes determining whether to further split a CU based on coding efficiency and quality. Coding efficiency and quality are determined using residual pixels values of a CU. In other words, to determine the efficiency and quality of the coding, the video in Sjoberg . . . is encoded. This involves generating and using residuals. Hence, by disclosing that CUs are further split based on coding efficiency and quality, Sjoberg . . . is disclosing using residual pixel values, even though the reference does not explicitly use the term. Reply Br. 2 (emphasis added); see App. Br. 6–7. Thus, because Sjoberg determines the efficiency and quality of the coding, Appellant concludes that Sjoberg must code all pixels, including residual pixels. Appellant contends that “[a]t the very least, the Examiner erred by failing to establish a prima facie case of inherency that Sjoberg . . . necessarily does not use residual pixel values.” Reply Br. 2 (italics added). Nevertheless, “a reference need not state a feature’s absence in order to disclose a negative limitation.” AC Techs. S.A. v. Amazon.com, Inc., 912 Appeal 2018-007268 Application 14/083,423 12 F.3d 1358, 1367 (Fed. Cir. 2019) (citing Sud-Chemie, Inc. v. Multisorb Techs., Inc., 554 F.3d 1001, 1004–05 (Fed. Cir. 2009) (affirming finding that reference disclosed “uncoated” film where it did not describe the film as coated and did not suggest necessity of coatings)). Thus, we read Sjoberg’s disclosure broadly; and we determine Sjoberg does not state that it does or does not use residual pixels, Appellant provides insufficient evidence to demonstrate that Sjoberg only describes coding using residual pixels, and the Examiner was not required to show the Sjoberg affirmatively teaches that residual pixel are not used for coding in order to establish a prima facie case of obviousness. Although the disputed limitations of claim 11 and claim 1 differ somewhat in scope, they are substantially similar, and Appellant relies on substantially the same arguments to challenge the rejection of both claims. Compare App. Br. 5–8; Reply Br. 2–8 with App. Br. 8–11; Reply Br. 9. Consequently, on this record, we are not persuaded that the Examiner erred in determining that the combined teachings of Sjoberg and Hebel render independent claims 1 and 11 obvious, and we sustain that rejection. 2. Dependent Claims 2 and 3 Claim 2 recites “selecting the parent CU as a partition of the CU partitioning when the first statistical measure of the parent CU does not exceed a third threshold and the second statistical measure of the parent CU does not exceed a fourth threshold.” App. Br. 15 (Claims App’x) (emphases added). Thus, this claim recites hierarchical coding to lower layers. See Spec. ¶ 55, Fig. 6. Sjoberg teaches that “[t]he LCUs . . . may be split into smaller coding units (CUs), which in turn may be split hierarchically in a quadtree fashion down to a smallest coding unit (SCU)” Sjoberg ¶ 5; see id. Appeal 2018-007268 Application 14/083,423 13 ¶ 48 (“As is well known in the art, a quadtree is a tree data structure in which each internal node has exactly four children.” (emphasis added)). Thus, the Examiner determines that Sjoberg teaches or suggests this limitation. Final Act. 8. Appellant contends that no mention is made in the applied references individually or in the Examiner’s proposed combination of references of four different thresholds – two of which are used to select whether to use the LCU as the CU partitioning and two more of which are used to select a parent CU as a partition for one or more lower layers. App. Br. 12. Further, Appellant contends, because Sjoberg does not teach computing even a first statistical measure based on the homogeneity or detail of the image area, Sjoberg and Hebel cannot teach second, third, or fourth statistical measures. Reply Br. 6. The Examiner asserts, however, Sjoberg and Hebel teach splitting the image into a plurality of CUs in a quadtree fashion based on the coding process. Ans. 17 (citing Sjoberg ¶ 51). Moreover, referring to Sjoberg’s Figures 3A and 3B, Sjoberg teaches at least three hierarchical layers comprising quadtrees, including four CUs 20, up to sixteen CUs 30, and up to sixty-four CUs 40, respectively. See Sjoberg ¶ 80. From these teachings, the Examiner concludes that Sjoberg teaches up to six statistical measures over at least three hierarchical layers, two per hierarchical layer. Ans. 17. Because, for the reasons given above, we are not persuaded that the Examiner erred with respect to the application of the combined teachings of Sjoberg and Hebel to independent claim 1 and because both Sjoberg and Hebel teach coding at a plurality of layers (see Sjoberg ¶¶ 5, 48; Hebel ¶ 14), Appeal 2018-007268 Application 14/083,423 14 we are persuaded the Examiner has shown that claim 2 is rendered obvious by the combined teachings of Sjoberg and Hebel. Claim 3 depends from claim 2 and recites “selecting child CUs of the parent CU in at least one lower layer of the CU hierarchy as part of the CU partitioning when the first statistical measure of the parent CU exceeds the third threshold and/or the second statistical measure of the parent CU exceeds the fourth threshold.” Id. at 16 (emphases added). This limitation is similar to the second “selecting” limitation of claim 1, discussed above. The Examiner relies on substantially the same teachings of Sjoberg applied to claim 2, to demonstrate that claim 3 is taught by the combined teachings of Sjoberg and Hebel. Final Act. 8. Appellant relies on the contentions presented with respect to claims 1 and 2 to challenge the rejection of claim 3. App. Br. 12–13; Reply Br. 7–8. For the reasons given above, we are not persuaded by these contentions that the Examiner erred in rejecting claim 3 over the combined teachings of Sjoberg and Hebel. Thus, on this record, we also sustain the Examiner’s rejection of claims 2 and 3 as rendered obvious over the combined teachings of Sjoberg and Hebel. CONCLUSION (1) The Examiner did not err in rejecting claims 1–3 and 11 as rendered obvious over the combined teachings of Sjoberg and Hebel. (2) Thus, claims 1–3 and 11 are unpatentable. DECISION We affirm the Examiner’s rejection of claims 1–3 and 11. No time period for taking any subsequent action in connection with Appeal 2018-007268 Application 14/083,423 15 this appeal may be extended under 37 C.F.R. § 1.136(a)(1)(iv). AFFIRMED Copy with citationCopy as parenthetical citation