Ex Parte Zhou et alDownload PDFPatent Trial and Appeal BoardAug 21, 201813463547 (P.T.A.B. Aug. 21, 2018) Copy Citation UNITED STA TES p A TENT AND TRADEMARK OFFICE APPLICATION NO. FILING DATE 13/463,547 05/03/2012 139020 7590 08/23/2018 BAKER & HOSTETLER LLP / Apple Inc. WASHINGTON SQUARE, SUITE 1100 1050 CONNECTICUT A VE. N.W. WASHINGTON, DC 20036-5304 FIRST NAMED INVENTOR Xiaosong Zhou 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 ATTORNEY DOCKET NO. CONFIRMATION NO. 082438.023801 5021 EXAMINER MCINNISH, KEVIN K ART UNIT PAPER NUMBER 2483 NOTIFICATION DATE DELIVERY MODE 08/23/2018 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): eofficemonitor@bakerlaw.com PTOL-90A (Rev. 04/07) UNITED STATES PATENT AND TRADEMARK OFFICE BEFORE THE PATENT TRIAL AND APPEAL BOARD Ex parte XIAOSONG ZHOU, JAMES OLIVER NORMILE, and HSI-JUNG WU Appeal2018-001033 Application 13/463,547 Technology Center 2400 Before ALLEN R. MacDONALD, MICHAEL M. BARRY, and MICHAEL J. ENGLE, Administrative Patent Judges. MacDONALD, Administrative Patent Judge. DECISION ON APPEAL 1 1 Appellants indicate the real party in interest is Apple Inc. App. Br. 3. Appeal2018-001033 Application 13/463,547 STATEMENT OF CASE Appellants appeal under 35 U.S.C. § 134(a) from a final rejection of claims 1-9, 11-25, 27, 29--45, 47--49, 52, 55, 58, 61, 64, and 66. The Examiner objects to claims 50, 51, 53, 54, 56, 57, 59, 60, 62, 63, 65, and 67. Final Act. 1. The Examiner indicates claims 68 and 69 are allowable. Id. Appellants have cancelled claims 10, 26, 28, and 46. App. Br. 24, 26, 28. We have jurisdiction under 35 U.S.C. § 6(b). We AFFIRM. Illustrative Claims Illustrative claims 1, 2, and 31, under appeal read as follows ( emphasis, formatting, and bracketed material added): 1. An encoder apparatus, comprising: [A.] a coding engine that performs predictive coding of input video; [B.] a decoder that decodes coded reference pictures generated by the coding engine; [C.] a reference picture cache that stores the decoded reference pictures; [D.] a patch cache that stores prediction patches generated from sources other than the decoded reference pictures from a current coding session; and [E.] a prediction reference selector that searches for a prediction reference from within the reference picture cache and, if the prediction reference is not found in the reference picture cache, the prediction reference selector searches for the prediction reference from within the patch cache. 2 2 Appellants' method claim 19 recites a similar conditional limitation of "if the prediction reference is not found." App. Br. 25. However, the construction of method claim 19 is significantly broader as its broadest reasonable interpretation encompasses an instance where the conditional limitation is not performed. See Ex parte Schulhauser, No. 2013-007847, 2016 WL 6277792, at *4 (PT AB April 28, 2016) (precedential) (holding 2 Appeal2018-001033 Application 13/463,547 2. The encoder of claim 1, wherein the stored prediction patches are derived from an identification of a predetermined type of object. 31. The method of claim 19, further comprising, when a new video coding session is created, transmitting contents of the patch cache to the channel. Rejections A. The Examiner rejected claims 1-7, 9, 11-25, 27, 29--44, 48, and 49 under 35 U.S.C. § I03(a) as being unpatentable over the combination of Dachiku et al. (US 5,592,228; issued Jan. 7, 1997) and Cosman et al. (US 2006/0098738 Al; pub. May 11, 2006). Appellants argue claim 1. App. Br. 10-17. Appellants argue claims 11, 19, 32, 43, and 44 solely by reference to the arguments presented for claim 1. App. Br. 17-18. Therefore, the rejection of claims 11, 19, 32, 43, and 44 turns on our decision as to claim 1. Claims 5, 7, 9, 15, 16, 18, 25, 27, 29, 33-37, 39, 48, and 49, depend from these claims; and Appellants do not argued them separately. We select claim 1 as representative. See 37 C.F.R. § 4I.37(c)(l)(iv). All the claims listed here stand or fall with claim 1. Appellants argue claims 2, 12, 20, and 23 as a group. 3 App. Br. 18- 19. We select claim 2 as representative. See 37 C.F.R. § 4I.37(c)(l)(iv). "[t]he Examiner did not need to present evidence of the obviousness of the remaining method steps of claim 1 that are not required to be performed under a broadest reasonable interpretation of the claim ... "). 3 "Under each heading identifying the ground of rejection being contested, any claim(s) argued separately or as a subgroup shall be argued under a separate subheading that identifies the claim(s) by number." 37 C.F.R. § 4I.37(c)(l)(iv). 3 Appeal2018-001033 Application 13/463,547 Claims 3, 4, 13, 14, 21, and 22, depend from these claims; and Appellants do not argue them separately. Therefore, they stand or fall with claim 2. Appellants argue claims 6, 17, 24, and 38 as a group. App. Br. 19-20. We select claim 6 as representative. See 37 C.F.R. § 4I.37(c)(l)(iv). Appellants argue claims 30, 31, and 40-42 as a group. App. Br. 20- 22. We select claim 31 as representative. See 37 C.F.R. § 4I.37(c)(l)(iv). Except for our ultimate decision, we do not discuss the § 103 rejection of claims 3-5, 7, 9, 11-25, 27, 29, 30, 32--44, 48, and49 further herein. B. The Examiner rejects claims 8, 45, 47, 52, 55, 58, 61, 64, and 66 under 35 U.S.C. § 103(a) as being unpatentable over the combination of Dachiku and Cosman in various combinations with other references. Final Act. 11-14. Appellants do not present arguments for claims 8, 45, 47, 52, 55, 58, 61, 64, and 66. Thus, the rejections of these claims tum on our decision as to claim 1. Except for our ultimate decision, we do not discuss the § 103 rejections of claims 8, 45, 47, 52, 55, 58, 61, 64, and 66 further herein. Issues on Appeal Did the Examiner err in rejecting claims 1, 2, 6, and 30 as being obvious? ANALYSIS We have reviewed the Examiner's rejections in light of Appellants' arguments (Appeal Brief and Reply Brief) that the Examiner has erred. We 4 Appeal2018-001033 Application 13/463,547 disagree with Appellants. We concur with the conclusions reached by the Examiner. We highlight the following points. A. Appellants raise the following argument in contending that the Examiner erred in rejecting claim 1 under 35 U.S.C. § 103(a). The Examiner states that Dachiku teaches a prediction reference cache and that Cosman teaches a patch cache comprising prediction patches generated from sources other than decoded reference pictures from a current coding session. Office Action at 7. The examiner is incorrect. There are several flaws with the Examiner's analysis. First, Cosman does not teach "a patch cache that stores prediction patches generated from sources other than the decoded reference pictures from a current coding session." Cosman teaches storage of reference frames that are drawn from a common coding session, not from outside a coding session, as claimed. App. Br. 10-11. The Examiner' [ s] rejection on this issue is based on a tortured construction of the term "coding session." In the Examiner's view, any change to a connection used to transmit the video being coded causes a new coding session to be created. The Examiner provided no evidence to support his construction. And neither Cosman nor Applicants' specification provide any disclosure to justify the Examiner's construction. The Examiner's rationale is wrong. The Oxford English Dictionary defines the word "session" as "A period of time given to or set aside for the pursuit of a particular activity." App. Br. 12. The Examiner presents the following response to Appellants' above argument. Examiner disagrees and notes there exists ample evidence to the contrary in Appellant's previous remarks, including Appellant's use of an entirely different definition of the term 5 Appeal2018-001033 Application 13/463,547 "session", the claimed invention itself, and the disclosure of the cited prior art that supports the Office's position that "coding session" of claim 1 depends directly on the connection. Referring to Appellant filed remarks received on 10/19/2016, Appellant points to http://searchsoa.techtarget.com/ definition/session to establish that the definition of session is "a series of interactions between two communication end points that occur during the span of a single connection .... and terminates when the connection is ended." (Remarks filed 10/19/2016, page 14). Yet, on page 12 of the brief filed 07/03/2017, Appellant turns to the Oxford English Dictionary. Examiner has afforded the claimed invention the previous definition as is evidenced by the prosecution history. Coincidentally, it is now this very definition that Appellant argues against. Ans. 5-6. 4 We agree with the Examiner that contrary to Appellants' Appeal Brief argument, the definition of "session" previously provided by Appellants is an appropriate definition for the term "session" in claim 1 (as previously argued by Appellants). We also agree with the Examiner that based on this 4 At page 14 of remarks filed on 10/19/2016, Appellants state: [T]he concept of a session is well understood in in [sic] computer science fields. Authoritative sources describe it as such: In telecommunication, a session is a series of interactions between two communication end points that occur during the span of a single connection. Typically, one end point requests a connection with another specified end point and if that end point replies agreeing to the connection, the end points take turns exchanging commands and data ("talking to each other"). The session begins when the connection is established at both ends and terminates when the connection is ended. See, http:// searchsoa. techtarget. com/ definition/ session. 6 Appeal2018-001033 Application 13/463,547 previously provided definition, Cosman discloses patches generated from sources other than the current coding session. 5 B. Appellants also raise the following argument in contending that the Examiner erred in rejecting claim 1 under 35 U.S.C. § 103(a). Cosman is silent ... regarding any order of searching for the short-term and long term frames in the frame memories. Moreover, because both of Cosman' s short-term frame and long- term frame are needed to perform the comparison, there is no suggestion that the order of searching the frame memories is relevant to Cosman' s encoder, since the same comparison would be performed regardless of the order of search. Accordingly, Cosman does not teach giving search precedence to either the short-term or long-term frame memory. App. Br. 14 (Appellants' emphasis omitted; Panel emphasis added). Contrary to Appellants' argument, Cosman is not silent regarding the order of searching. Rather, Cosman explicitly states: In embodiments of the invention, we assume that the encoder codes according to standard procedure, namely the previous frame is used as the reference frame. However, the decoder can make use of an additional reference frame (the long term past 5 Because Appellants and the Examiner have treated as meaningful the claim 1 limitation of "generated from sources other than the decoded reference pictures from a current coding session" ( and as discussed above, we agree with the Examiner's selected definition), we follow their lead for purposes of this decision. However, had we agreed with the Appellants' argued definition, our conclusion would have remained the same. This argued limitation is directed to an intangible property of the stored data, i.e., its source. We conclude that such a limitation does not serve to structurally distinguish over an otherwise obvious prior art structure. Also, it unclear how the objected to encoder/decoder claims 50, 51, 53, and 54 (and rejected claims 2-7 and 12-18) distinguish over the prior art as they similarly recite only an intangible property of the stored data. 7 Appeal2018-001033 Application 13/463,547 high quality frame) as well as the previous frame. Instead of concealing only when there is a loss, the decoder might use concealment when there is a very poor quality macroblock that is received that could be replaced with a block of higher quality by using the past high quality reference frame. This could be called improvement rather than concealment, but it can be considered a variant on temporal concealment. In essence, when the quality is poor enough, we might choose to consider that a block has been lost even when in fact it hasn't been, and we can invoke a loss concealment approach in which the high quality past frame is used to replace the lost block. Cosman ,r 33 (emphasis added). Thus, contrary to Appellants' argument, Cosman teaches giving precedence to the short-term reference frame over the long-term reference frame, and only using "concealment" (i.e., using an older "high quality reference frame") either (A) "when there is a loss" or (B) "when the quality [of the previous frame] is poor enough." This teaches or suggests using a patch cache "if the prediction reference is not found in the reference picture cache," as recited in claim 1. 6 C. Appellants further raise the following argument in contending that the Examiner erred in rejecting claim 1 under 35 U.S.C. § 103(a). The present rejection also should be reversed because the prior art provides no motivation or other reason to combine the teachings of the references, which propose different solutions to very different problems in coding systems. 6 Although Appellants' claim 1 recites "if the prediction reference is not found," Appellants' Specification discloses this "not found" only in the context of "[i]f an adequate prediction reference is not found." Spec. ,r 30 ( emphasis added); see also App. Br. 6 ( citing for support of this limitation only Spec. ,r,r 17, 18, 29-34, Figs. 2, 4). Paragraph 33 of Cosman is similarly directed to if the prediction reference is inadequate (i.e., yields "very poor quality"), then the past high quality reference frame is used. 8 Appeal2018-001033 Application 13/463,547 Dachiku discloses a video coding apparatus that incorporates a moving object analysis stage to increase the accuracy of the coding video signals that containing a variety of motion types, such as rotation, elongation, or contraction. Dachiku at 4: 15-20. Cosman, on the other hand, depicts an encoder that employs what is generally conventional inter-frame coding, extended to perform the inter-frame coding using long-term reference frames under certain conditions, such as when the quality of the input pictures deteriorates due to factors such as bandwidth fluctuations. Cosman at [0012] and [0066]. Because Dachiku and Cosman address different solutions to different problems in coding systems, it would not have been obvious to combine their teachings in the manner proposed by the Examiner. The Examiner seems to propose using Cosman 's long-term reference frames in Dachiku 's object motion analysis. However, Dachiku 's motion analysis is dependent on taking "time-wise" differences between input frames .... Taking the time-wise difference using a long-term frame, as suggested by the Examiner, would be expected to reduce the accuracy of the moving object analysis, .... . . . . Here, combining Dachiku and Cosman would not have yielded predictable results, since the proposed combination would reduce the accuracy of Dachiku 's moving object analysis. App. Br. 15-16 (emphasis added). We are not persuaded by Appellants' argument. As Appellants acknowledge, Cosman is directed to an inter-frame coding technique applicable to certain conditions where quality deteriorates due to factors such as bandwidth fluctuations. First, Appellants' "no motivation or other reason to combine" argument overlooks that Dachiku is also inter-frame coding. Dachiku 1:8-10; see also Ans. 7-8. Second, Appellants' "reduced accuracy" argument is premised on applying Cosman's long-term reference 9 Appeal2018-001033 Application 13/463,547 frames to Dachiku during normal conditions (scenario A) as opposed to certain conditions (scenario B) where quality deteriorates in Dachiku due to factors such as bandwidth fluctuations. Even if we conclude Appellants are correct that accuracy is reduced if the long-term reference frames technique of Cosman is used during normal conditions, normal conditions is not the scenario under which Cosman teaches to use long-term reference frames. Contrary to Appellants' argument, we conclude that combining Dachiku and Cosman to operate under deteriorated quality conditions, for which Cosman is designed, would yield predictable results. Third, the Examiner also provides a reason to combine "in order to reduce computation complexity." Final Act. 7 ( citing Cosman ,r 25); Cosman ,r 25 ( explaining why its approach "may be less complex than in the prior art"). D. Appellants raise the following argument in contending that the Examiner erred in rejecting claim 2 under 35 U.S.C. § 103(a). The Examiner admits that Dachiku does not disclose prediction patches or a patch cache but somehow concludes that Dachiku 's object motion analysis teaches how such patches would be derived. Office Action at 7. The Examiner is wrong. Dachiku discloses a video coding apparatus that incorporates a moving object analysis stage to increase the accuracy of the coding video signals that containing a variety of motion types, such as rotation, elongation, or contraction. Dachiku also describes that the polygonal regions around such features may be reduced in size, in order to more accurately estimate finer motions around these features .... The teachings of Dachiku thus relate to motion estimation, not to deriving data that will be stored in any cache. Dachiku 's polygonal regions do not become prediction data that is stored in any cache. Instead, Dachiku 's system populates its cache with 10 Appeal2018-001033 Application 13/463,547 decoded data from within a coding session in the ordinary manner. App. Br. 18-19 (emphasis added). We are not persuaded by Appellants' argument. We see no relevance to Appellants' claim 2 discussion of "variety of motion types, such as rotation, elongation, or contraction." This was not the basis of the rejection of claim 2. Rather, in rejecting claim 1, the Examiner relied on a combination of Dachiku and Cosman which, as discussed above, we agree renders obvious claim 1. Claim 2 depends from claim 1 and adds identification of a predetermined type of object. The Examiner correctly points out that Dachiku discloses that video encoding can be improved by including (1) identifying a predetermined type of object, i.e., including a feature point extractor and a human object detector (Dachiku 10:43--44), and (2) reducing the size of the area of video processing to deal with finer detail (Dachiku 10:47--49). We agree with the Examiner's conclusion that it would have been obvious to apply this disclosed video encoding improvement (which teaches the additional limitation of claim 2) to the prediction patch processing of claim 1. E. Appellants raise the following argument in contending that the Examiner erred in rejecting claim 6 under 35 U.S.C. § 103(a). The rejection[] of claim[] 6 . . . should be reversed[] because it would not have been obvious to predefine prediction patches prior to the onset of a video coding session. The Examiner asserts that these claims would be obvious in view of Dachiku and Cosman, based on Cosman 's disclosure of a long-term reference frame held in a frame memory. Office 11 Appeal2018-001033 Application 13/463,547 Action at 8. However, the Examiner is wrong. Cosman 's long- term reference frames are merely reference frames retained from earlier points in the same coding session. As such, they are not predefined prior to the onset of the coding session. App. Br. 19-20 ( emphasis added). Appellants again present the same coding session argument they make for claim 1. Again, we are not persuaded for the reasons already discussed above. F. Appellants raise the following argument in contending that the Examiner erred in rejecting claim 31 under 35 U.S.C. § 103(a). Claim[] 31 ... recite[ s] transmitting or receiving patches from the patch cache when the session is established. The Examiner wrongly purports that this is ... taught by the combination of Dachiku and Cosman, in view of Dachiku's description of receiving the coded video data at an input terminal of a decoding apparatus. Office Action at 10-11, citing Dachiku at 12:1-29. However, the cited portion of Dachiku - which merely describes coded data being sent to a decoding apparatus - teaches nothing about transmitting the contents of a patch cache when establishing a decoding session. App. Br. 20-21. We are unpersuaded by Appellants' argument. As Appellants acknowledge, Dachiku at 12: 1-29 teaches coded data being sent to a decoding apparatus, or as the Examiner finds, Dachiku discloses "transmitting contents ( e.g. frame and patch data) across a transmission channel." Final Act. 11. Although Appellants argue that the Examiner erred in finding Dachiku teaches "transmitting the contents of a patch cache when 12 Appeal2018-001033 Application 13/463,547 establishing a decoding session," our review determines the Examiner did not make any such finding. Rather, looking at the rejection of claim 31 as a whole (including the rejection of claim 19 from which it depends), we determine that Cosman at paragraphs 23-25 ( cited by the Examiner in rejecting claim 19) discloses examples of the timing of new patch data creation and thus the needed use of the transmission channel to transmit that new patch data. Cosman at paragraph 23 states operating conditions including "a change in the quality of the connection over which video is being received." An artisan would have understood "a change in the quality" to include both decreasing quality and increasing quality. Cosman also states: When a recent reference frame is higher quality than the past reference frame, it is then retained as the past high quality reference frame in the long term reference buffer, and the previous past high quality reference frame is discarded." Cosman ,r 24. An artisan would have understood that when a new higher quality coding session is established, a new high quality reference frame (a patch) is also established, which then requires it be transmitted to the decoding apparatus as taught by Dachiku. We conclude nothing more than this is required by claim 31. CONCLUSIONS (1) The Examiner has not erred in rejecting claims 1-9, 11-25, 27, 29--45, 47--49, 52, 55, 58, 61, 64, and 66 as being unpatentable under 35 U.S.C. § 103(a). (2) Claims 1-9, 11-25, 27, 29--45, 47--49, 52, 55, 58, 61, 64, and 66 are not patentable. 13 Appeal2018-001033 Application 13/463,547 DECISION The Examiner's rejections of claims 1-9, 11-25, 27, 29--45, 47--49, 52, 55, 58, 61, 64, and 66 are affirmed. No time period for taking any subsequent action in connection with this appeal may be extended under 37 C.F.R. § 1.136(a)(l )(iv). AFFIRMED 14 Copy with citationCopy as parenthetical citation