DETAILED ACTION
Continued Examination Under 37 CFR 1.114
A request for continued examination under 37 CFR 1.114, including the fee set forth in 37 CFR 1.17(e), was filed in this application after final rejection. Since this application is eligible for continued examination under 37 CFR 1.114, and the fee set forth in 37 CFR 1.17(e) has been timely paid, the finality of the previous Office action has been withdrawn pursuant to 37 CFR 1.114. Applicant's submission filed on 04/13/2026 has been entered.
Response to Arguments
Examiner incorporates herein previous Responses to Arguments.
On page 9 of the Remarks, Applicant contends, Chiang does not disclose or suggest any mechanism for determining the number of non-separable transform kernels included in the plurality of transform kernels for the primary transform.” Examiner finds Applicant does not argue that which is claimed. The plurality of transforms to choose from is interpreted to include secondary transforms because Applicant has not claimed otherwise. Furthermore, as explained in the preceding Office Action and with respect to a feature removed from claim 1 and included now with claim 12, Examiner has applied an interpretation to the claimed subject matter regarding applying a non-separable transform to the primary transform as markedly different from a recitation that would recite applying a non-separable transform as the primary transform. Examiner’s interpretation is consistent with the prior art, the level of skill in the art, the semantics of the English language (i.e. plain meaning doctrine), and Applicant’s published paragraph [0126]. Furthermore, as explained under the Conclusion Section of this Office Action, using a non-separable transform as a primary transform is not novel (see Zhao (US 2021/0160519 A1), ¶ 0179 and Zhao (US 2022/0078423 A1), ¶ 0218 and Siekmann (US 2021/0084301 A1), ¶¶ 0453 and 0468). To expedite prosecution, in addition to providing additional prior art references that demonstrate that the skilled artisan would have interpreted Egilmez’s teaching of a primary DCT transform as covering both separable and non-separable DCT, the rejection now additionally relies on the teachings of Seregin, ¶ 0088, which concretely explains that both primary and secondary transforms in this art are envisaged to include an index that signals any separable or non-separable transform, whether as a primary or secondary transform. Applicant’s specific argument that a position of a last coefficient can determine whether only one primary transform is applied does not seem to be squarely supported by Applicant’s Specification, at least upon the review of the Examiner. Examiner requests Applicant’s assistance in identifying where in the Specification Applicant finds support for arguing the averred interpretation. Absent further clarification in the claim and in the record, Examiner is unpersuaded of error.
Other claims are not argued separately. Remarks, 10–11.
Claim Rejections - 35 USC § 103
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102 of this title, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
Claims 1, 2, 4, 5, 12–14, 16, and 20 are rejected under 35 U.S.C. 103 as being unpatentable over Egilmez (US 2020/0366937 A1), Chiang (US 2023/0130131 A1), and Seregin (US 2018/0367814 A1).
Regarding claim 1, the combination of Egilmez, Chiang, and Seregin teaches or suggests an image decoding method performed by a decoding apparatus (Egilmez, Abstract: teaches a decoder), the method comprising: obtaining image information through a bitstream (Egilmez, ¶ 0005: teaches obtaining encoded video data through a bitstream), wherein the image information includes residual information and transform index information (Egilmez, ¶ 0047: teaches the encoded video data includes residual data and coding decision information in the form of syntax elements); deriving transform coefficients for a current block based on the residual information (Egilmez, ¶ 0070: teaches transform coefficients represent residual data); deriving residual samples for the current block by performing a primary transform using a transform kernel related to the transform index information among a plurality of transform kernels based on the transform coefficients (Egilmez, ¶ 0096: teaches primary and secondary transforms can be applied to a residual block; Egilmez, ¶ 0121: teaches both LFNST and MTS transforms, which the skilled artisan knows are signaled using indexes, lfnst_idx and mst_idx; Chiang, ¶ 0202: teaches LFNST and MTS indexes being signaled to indicate whether either is disabled and, if not, what transform to use; While not viewed necessary to sustain the rejection, and in response to Applicant arguing the prior art did not teach a flag indicating selection of a non-separable primary transform, Seregin, ¶ 0088, unequivocally teaches EMT/MTS utilizing a plurality of transform kernels to include both separable and non-separable primary transforms); and generating reconstructed samples for the current block based on the residual samples, wherein the plurality of transform kernels includes a separable transform kernel and a non-separable transform kernel (see above wherein it is explained both Egilmez and Chiang teach LFNST and MTS; Egilmez, ¶ 0097: teaches LFNSTs are non-separable and MTS transforms may be one or more separable transforms), wherein a number of non-separable transform kernels included in the plurality of transform kernels is determined based on position information of a last significant transform coefficient in the current block (Egilmez, ¶¶ 0007 and 0148: teaches that when the position of the last significant coefficient (known to be included with residual information; see Applicant’s original claim 2 confirming this interpretation) indicates being in a zero-out region known to be associated with certain LFNSTs, the decoder can infer the value of the LFNST index, perhaps inferring an LFNST index of zero (meaning LFNST disabled) if the last coefficient is in a position that no LFNST can support; Egilmez, ¶¶ 0148 and 0150: explain that because LFNSTs can have normative zero-out regions, when there is a last significant coefficient in one of those regions, its presence serves to rule-out that particular LFNST since the LFNST’s zero-out region is incompatible with there being a significant (i.e. non-zero) coefficient present within that zero-out region; Given these teachings of Egilmez, it would have been obvious for the skilled artisan to arrive at Applicant’s claimed feature that a certain number of non-separable transform kernels (i.e. LFNSTs) can be ruled-out, i.e. pruned, from the total set of available LFNSTs (or primary/secondary transform combinations) based on the position of the last non-zero coefficient; Chiang, ¶ 0015: teaches “conditioning of the LFNST index on the last-significant position”; see also Chiang, ¶ 0015: teaches disabling LFNST when the last significant coefficient is less than a threshold (which also reduces the number of primary/secondary transform combinations)), wherein based on a value of a first variable being less than or equal to a predetermined first threshold (Examiner interprets this feature as saying the two-dimensional block can be scanned into a one-dimensional array of coefficients and that the position of the last coefficient is signaled with reference to the one-dimensional serialized data element, wherein top-left is 0 and the scan moves towards the bottom-right; Egilmez, ¶ 0007: teaches the last non-zero coefficient is referenced in terms of the sequenced/scanned block in scanning order; Egilmez, ¶ 0065: teaches scanning the coefficients produces a one-dimensional (serialized) vector from the two-dimensional matrix; Egilmez, ¶ 0156: teaches the position of the last significant coefficient can be treated either in two-dimensions using (X,Y)-coordinate or by index value for a one-dimensional list), the plurality of transform kernels includes one non-separable transform kernel, wherein based on the value of the first variable exceeding the first threshold, the plurality of transform kernels includes a plurality of non-separable transform kernels, and wherein the first variable represents one-dimensional information of the last significant coefficient in the current block (Chiang, ¶ 0015: teaches disabling LFNST when the last significant coefficient is less than a threshold; See also Jung ‘070 cited under Conclusion Section of this Office Action; Chiang, then, teaches the skilled artisan that only the primary transform would be used when the scan index of the last significant coefficient is less than the threshold and teaches when the scan index of the last sig coeff exceeds the threshold, both primary and secondary transforms are available).
One of ordinary skill in the art, before the effective filing date of the claimed invention, would have been motivated to combine the elements taught by Egilmez, with those of Chiang, because both references are drawn to the same field of endeavor and because Chiang is merely explaining what the skilled artisan already knew, that Egilmez’s MTS and LFNST signaling included syntax elements for the MTS and LFNST indexes for primary and secondary transforms, respectively. Therefore, the combination is a mere combination of prior art elements, according to known methods, to yield a predictable result. This rationale applies to all combinations of Egilmez and Chiang used in this Office Action unless otherwise noted.
One of ordinary skill in the art, before the effective filing date of the claimed invention, would have been motivated to combine the elements taught by Egilmez and Chiang, with those of Seregin, because all three references are drawn to the same field of endeavor and because Seregin is merely explaining what the skilled artisan already knew, that Egilmez’s MTS and DCT for the primary transform could be signaled using a flag and could include both separable and non-separable transforms as primary transforms. Therefore, the combination is a mere combination of prior art elements, according to known methods, to yield a predictable result. This rationale applies to all combinations of Egilmez, Chiang, and Seregin used in this Office Action unless otherwise noted.
Regarding claim 2, the combination of Egilmez, Chiang, and Seregin teaches or suggests the image decoding method of claim 1, wherein the residual information includes first syntax elements related to position information of a last significant coefficient (Egilmez, ¶¶ 0007 and 0148: teaches that when the position of the last significant coefficient indicates being in a zero-out region known to be associated with certain LFNSTs, the decoder can infer the value of the LFNST index, perhaps inferring an LFNST index of zero (meaning LFNST disabled) if the last coefficient is in a position that no LFNST can support; Egilmez, ¶¶ 0148 and 0150: explain that because LFNSTs can have normative zero-out regions, when there is a last coefficient in one of those regions, its presence serves to rule-out that particular LFNST since the LFNST’s zero-out region is incompatible with there being a significant (i.e. non-zero) coefficient present within the zero-out region; Given these teachings of Egilmez, it would have been obvious for the skilled artisan to arrive at Applicant’s claimed feature that a certain number (or all) of non-separable transform kernels (i.e. LFNSTs) can be ruled-out, i.e. pruned, from the total set of available LFNSTs based on the position of the last non-zero coefficient; Chiang, ¶ 0015: teaches “conditioning of the LFNST index on the last-significant position”), wherein the first variable is derived based on the first syntax elements (Examiner interprets this feature as saying the two-dimensional block is scanned into a one-dimensional array of coefficients and that the position of the last coefficient is signaled with reference to the one-dimensional serialized data element, wherein top-left is 0 and the scan moves towards the bottom-right; Egilmez, ¶ 0007: teaches the last non-zero coefficient is referenced in terms of the sequenced/scanned block in scanning order; Egilmez, ¶ 0065: teaches scanning the coefficients produces a one-dimensional (serialized) vector from the two-dimensional matrix; Egilmez, ¶ 0156: teaches the position of the last significant coefficient can be treated either in two-dimensions using (X,Y)-coordinate or by index value for a one-dimensional list), and wherein the first variable has a value of 0 for a top-left sample position of the current block, and the first variable has a larger value for samples located in a bottom-right based on the top-left sample position of the current block (Egilmez, ¶ 0097–0098: teaches what the skilled artisan already knows, which is that the low frequencies are top-left and proceed to higher frequencies toward the bottom-right; Egilmez, ¶ 0065: teaches the lower frequencies are at the front of the vector and the higher frequencies are at the back of the vector).
Regarding claim 4, the combination of Egilmez, Chiang, and Seregin teaches or suggests the image decoding method of claim 1, wherein based on a value of the first variable being less than or equal to a predetermined second threshold, the plurality of transform kernels include one non-separable transform kernel, wherein based on the value of the first variable exceeding the second threshold and being less than or equal to a predetermined third threshold, the plurality of transform kernels include M non-separable transform kernels, wherein based on the value of the first variable exceeding the third threshold, the plurality of transform kernels includes N non-separable transform kernels, and wherein M is an integer greater than 1, and N is an integer greater than M (Applicant’s published paragraph [0256] explains multiple ranges (i.e. multiple thresholds) can include looking for a DC-only coefficient utilizing a threshold of 1; Chiang, ¶ 0125: explains that LFNST DC Only syntax element can be determined based on the last significant coefficient position; Examiner notes the skilled artisan understands that the LFNST DC Only signal means a threshold of 1 as evidenced by Jung ‘070, ¶ 0290; Egilmez, ¶¶ 0007 and 0148: teaches that when the position of the last significant coefficient (known to be included with residual information; see Applicant’s original claim 2 confirming this interpretation) indicates being in a zero-out region known to be associated with certain LFNSTs, the decoder can infer the value of the LFNST index, perhaps inferring an LFNST index of zero (meaning LFNST disabled) if the last coefficient is in a position that no LFNST can support; Egilmez, ¶¶ 0148 and 0150: explain that because LFNSTs can have normative zero-out regions, when there is a last coefficient in one of those regions, its presence serves to rule-out that particular LFNST since the LFNST’s zero-out region is incompatible with there being a significant (i.e. non-zero) coefficient present within that zero-out region; Given these teachings of Egilmez, it would have been obvious for the skilled artisan to arrive at Applicant’s claimed feature that a certain number of non-separable transform kernels (i.e. LFNSTs) can be ruled-out, i.e. pruned, from the total set of available LFNSTs based on the position of the last non-zero coefficient; Chiang, ¶ 0015: teaches “conditioning of the LFNST index on the last-significant position”; see also Chiang, ¶ 0015: teaches disabling LFNST when the last significant coefficient is less than a threshold).
Regarding claim 5, the combination of Egilmez, Chiang, and Seregin teaches or suggests the image decoding method of claim 1, wherein based on a value of the first variable being 0, the transform index information is related to the separable transform kernel (Examiner interprets the first variable being 0 to indicate DC only; Chiang, ¶ 0015: teaches disabling LFNST when the last significant coefficient is less than a threshold, which means only MTS index and not LFNST index needs to be parsed, the MTS index typically being related to separable transform kernels; Egilmez, ¶ 0097: teaches LFNSTs are non-separable and MTS transforms may be one or more separable transforms).
Regarding claim 12, the combination of Egilmez, Chiang, and Seregin teaches or suggests the image decoding method of claim 1, wherein the image information further includes a flag representing whether the non-separable transform kernel is applied to the primary transform (This feature is interpreted consistent with Applicant’s published paragraph [0126], wherein it is explained the NSST (i.e. LFNST; see Egilmez, ¶ 0138) is applied to the primary transformed coefficients; Examiner notes this is how the art views the secondary transform; see e.g. Egilmez, ¶ 0063: teaching, similar to Applicant’s paragraph [0126], that MDNSST is applied following the first transform; Egilmez, ¶ 0071: teaches the LFNST index is also known as a flag), and wherein the residual information is parsed before the flag and the transform index information related to the non-separable transform kernel (The residual information in this art includes information on the last significant coefficient; See e.g. Chiang, Table 12: teaching information on the last significant coefficient is in the syntax table for residual coding; Egilmez, ¶¶ 0007 and 0148: teaches that when the position of the last significant coefficient indicates being in a zero-out region known to be associated with certain LFNSTs, the decoder can infer the value of the LFNST index, perhaps inferring an LFNST index of zero (meaning LFNST disabled) if the last coefficient is in a position that no LFNST can support; Egilmez, ¶¶ 0148 and 0150: explain that because LFNSTs can have normative zero-out regions, when there is a last coefficient in one of those regions, its presence serves to rule-out that particular LFNST since the LFNST’s zero-out region is incompatible with there being a significant (i.e. non-zero) coefficient present within the zero-out region; Chiang, ¶ 0015: teaches “conditioning of the LFNST index on the last-significant position”; Given these teachings of Egilmez and Chiang, it would have been obvious for the skilled artisan to arrive at Applicant’s claimed feature that the position of the last significant coefficient should be signaled before the LFNST index or MTS index since some (or all) of non-separable transform kernels (i.e. LFNSTs) can be ruled-out, i.e. pruned, from the total set of available LFNSTs based on the position of the last non-zero coefficient).
Claim 13 lists the same elements as claim 1, but is drawn to the corresponding encoding method rather than the decoding method. Therefore, the rationale for the rejection of claim 1 applies to the instant claim.
Claim 14 lists the same elements as claim 2, but is drawn to the corresponding encoding method rather than the decoding method. Therefore, the rationale for the rejection of claim 2 applies to the instant claim.
Claim 16 lists the same elements as claim 4, but is drawn to the corresponding encoding method rather than the decoding method. Therefore, the rationale for the rejection of claim 4 applies to the instant claim.
Claim 20 lists essentially the same elements as claim 1, but is drawn to further transmitting the encoded data to the decoder. Since the whole point of encoding is storage or transmission of the data, the rationale for the rejection of claim 1 applies to the instant claim.
Claim 6 is rejected under 35 U.S.C. 103 as being unpatentable over Egilmez, Chiang, Seregin, and Egilmez (US 2021/0211685 A1) (herein “Egilmez ‘685”).
Regarding claim 6, the combination of Egilmez, Chiang, Seregin, and Egilmez ‘685 teaches or suggests the image decoding method of claim 1, wherein based on a value of the first variable being 0, the transform index information is related to a predetermined transform kernel (Examiner notes that when the last significant position is the DC position, then the block is smooth; Chiang, ¶ 0009: teaches DCT2 is the default transform when no index is specified; Egilmez ‘685, ¶¶ 0127–0128: teaches that when only the DC coefficient is non-zero, neither the LFNST nor MTS index need to be signaled; Egilmez, ‘685, ¶ 0120: teaches the last significant coefficient being greater than a threshold indicates whether there is a DC coefficient).
One of ordinary skill in the art, before the effective filing date of the claimed invention, would have been motivated to combine the elements taught by Egilmez, Chiang, and Seregin, with those of Egilmez ‘685, because all four references are drawn to the same field of endeavor such that one wishing to practice transform index signaling based on position of last significant transform coefficient information would be led to their relevant teachings and because Egilmez ‘685 is merely explaining what the skilled artisan already knew, that MTS signaling for a DC only coefficient is not the purpose of extending the number of transform types available to code video blocks since there is no benefit to be gained and since bits can be saved by assuming the default transform mode in such circumstances. Therefore, the combination is a mere combination of prior art elements, according to known methods, to yield a predictable result. This rationale applies to all combinations of Egilmez, Chiang, Seregin, and Egilmez ‘685 used in this Office Action unless otherwise noted.
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure.
Jung (US 2021/0321136 A1) teaches that the decoder can decide whether to parse the MTS index, mts_idx, based on whether numSigCoeff is greater than 2 (¶¶ 0194–0197).
Jung (US 2021/0076070 A1) teaches processing syntax elements related to the secondary transform differently depending on the value of numSigCoeff (¶ 0148). It also teaches LFNST index may depend on last significant coefficient in scan order or numSigCoeff counter (¶ 0222). It also teaches, “Similar to when the number of significant coefficients is small, when the position (scan index) of the last significant coefficient in the scan order is small, coding efficiency due to the secondary transform may be low.” (¶ 0187). This publication was used to reject cancelled claims 7–11. Examiner notes that Jung, ¶ 0198: teaches that the skilled artisan was aware of using the numSigCoeff counter or the position of the last significant coefficient to infer whether to use certain transforms, which then links this claim with original claim 3, which uses LAST rather than numSigCoeff. See prosecution history.
Zhao (US 2021/0160519 A1) teaches the primary transform and secondary transform can both be non-separable (¶ 0179).
Siekmann (US 2021/0084301 A1) teaches separable DCT/DST primary transforms, non-separable primary only transforms, and primary+secondary transforms (¶ 0468).
Kanoh (US 2020/0137388 A1) teaches primary and secondary transforms can be either separable or non-separable (e.g. ¶ 0316).
Chiang (US 2022/0224898 A1) teaches only a DC coefficient means the secondary transform index can be skipped and teaches evaluating whether the last significant coefficient is above or below a threshold for determining the signaling of a transform index (¶ 0068).
Zhao (US 2022/0078423 A1) teaches that MTS signaling can be skipped when the position of the last significant coefficient is less than 1 (i.e. DC only) (¶ 0198) and teaches the primary transform kernels may be separable or non-separable (¶ 0218).
Kanumuri (US 2009/0046995 A1) teaches the DCT transform can be separable or non-separable (¶ 0064).
Pau (US 2014/0198995 A1) teaches separable and non-separable DCT (¶ 0069).
Any inquiry concerning this communication or earlier communications from the examiner should be directed to Michael J Hess whose telephone number is (571)270-7933. The examiner can normally be reached on Mon - Fri 9:00am-5:30pm.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, William Vaughn can be reached on (571)272-3922. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of an application may be obtained from the Patent Application Information Retrieval (PAIR) system. Status information for published applications may be obtained from either Private PAIR or Public PAIR. Status information for unpublished applications is available through Private PAIR only. For more information about the PAIR system, see http://pair-direct.uspto.gov. Should you have questions on access to the Private PAIR system, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative or access to the automated information system, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000.
MICHAEL J. HESS
Primary Examiner
Art Unit 2481
/MICHAEL J HESS/Primary Examiner, Art Unit 2481
Prosecution Insights