DETAILED ACTION
Claim Rejections - 35 USC § 112(b)
The following is a quotation of 35 U.S.C. 112(b):
(b) CONCLUSION.—The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the inventor or a joint inventor regards as the invention.
The following is a quotation of 35 U.S.C. 112 (pre-AIA ), second paragraph:
The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the applicant regards as his invention.
Claims 1, 13, 14 are rejected under 35 U.S.C. 112(b) or 35 U.S.C. 112 (pre-AIA ), second paragraph, as being indefinite for failing to particularly point out and distinctly claim the subject matter which the inventor or a joint inventor (or for applications subject to pre-AIA 35 U.S.C. 112, the applicant), regards as the invention.
The term “fast” in claim 1 is a relative term which renders the claim indefinite. The term “fast” is not defined by the claim, the specification does not provide a standard for ascertaining the requisite degree, and one of ordinary skill in the art would not be reasonably apprised of the scope of the invention. The delineation of speed into fast and not-fast has not been defined.
Claim 6 is rejected under 35 U.S.C. 112(b) or 35 U.S.C. 112 (pre-AIA ), second paragraph, as being indefinite for failing to particularly point out and distinctly claim the subject matter which the inventor or a joint inventor (or for applications subject to pre-AIA 35 U.S.C. 112, the applicant), regards as the invention. The phrase “contacted as a feature vector” is unclear. Is this supposed to be “concatenated”?
Claim Rejections - 35 USC § 103
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis (i.e., changing from AIA to pre-AIA ) for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
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, 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.
Claim(s) 1-2, 4-6, 8-10, 13-14 are rejected under 35 U.S.C. 103 as being unpatentable over Li (NPL: “CNN Based CU Partition Mode decision algorithm for HEVC Inter Coding,” IEEE 2018) in view of Taabane (NPL “Low Complexity Learning-Based QTMTT Partitioning Scheme for Inter Coding in VVC Encoder,” IEEE 2024) and Huang (NPL: “Block Partitioning Structure in the VVC Standard,” IEEE 2021).
Regarding Claim 1, Li (NPL: “CNN Based CU Partition Mode decision algorithm for HEVC Inter Coding,” IEEE 2018) discloses a computer-implemented method (platform is Huawei RH5885, Intel Xeon processor and RAM, Section 3) for making fast partitioning decision (fast CU mode, Abstract) on a block (CU, Abstract) of a frame (inherent) of a video (video, Introduction), the method comprising steps of:
…
b) for a first one of the plurality of partition modes (splitting of a 2Nx2N block into four NxN blocks, Section 2; i.e., quadtree in standards vernacular) and before attempting on a second one of the plurality of partition modes (split recursively until the minimum size 8x8, Section 1, or until the CNN returns HOMO, Section 2; i.e., only quadtree, not binary tree, ternary tree, or multi-type tree, in standards vernacular): predicting a RD (rate distortion) cost (
PNG
media_image1.png
24
120
media_image1.png
Greyscale
, Section 2.1) resulted from partitioning the block using the first one of partition modes (SubRDCi represents one of 4 sub-CUs’ IME cost, Section 2.1, therefore
PNG
media_image1.png
24
120
media_image1.png
Greyscale
represents the sum of all 4 sub CUs);
c) if the RD cost indicates an improved RD performance: (none of the three homogeneity conditions are satisfied, considering RD cost of the current CU and 4 sub-CUs, Section 2.1):
i) performing partitioning using the first one of the plurality of partition modes (SPLIT the 2Nx2N into four square smaller CUs, Section 1, Section 2);
ii) repeating Step b) (recursively, Section 1) for the first one of the plurality of partition modes (split the 2Nx2N into four square smaller CUs, Section 1; i.e., quadtree) based on the partitioned block in Step i) (the returned SPLIT pattern from the previous iteration, Section 2);
d) skipping the first one of the plurality of partition modes (otherwise, HOMO is returned by the CNN, Section 2) if the RD cost does not indicate an improved RD performance (at least one condition of homogeneity is satisfied, including RD cost of the current CU and four sub-CUs, Section 2.1).
Li does not disclose but Taabane (NPL “Low Complexity Learning-Based QTMTT Partitioning Scheme for Inter Coding in VVC Encoder,” IEEE 2024) a) identifying, for the block (for the CU, table 1), a plurality of partition modes (possible allowed partition modes based on size, Table 1, and partition mode constraints, Section II);
…
and e) following execution of Step d) (after the quadtree splitting is complete, Fig. 3, based on partitioning constraints in the encoding standard), predicting a RD cost (encoding efficiency, page 141096, left column ) resulted from partitioning the block using the second one of the plurality of partition modes (other allowable partitions such as binary tree, ternary tree, table 1, Fig. 3, each CU is evaluated for an eligible partition type, page 141096 left column, using partition type classifiers, page 141095 right column).
One of ordinary skill in the art before the application was filed would have been motivated to follow the quadtree partitioning of Li with the multitype tree partitioning of Taabane because Huang teaches that multi-type tree partitioning provides enhanced video compression over quad-tree-only partitioning for high resolution video (Section 1 second-to-last paragraph), providing higher quality video at faster speed to consumers, improving experience.
Regarding Claim 2, Li (NPL: “CNN Based CU Partition Mode decision algorithm for HEVC Inter Coding,” IEEE 2018) discloses the computer-implemented method of claim 1, wherein predicting the RD cost in Step b) is based on a plurality of features of the block (edge strength, quantization parameter and step, Section 2.1).
Regarding Claim 4, Li (NPL: “CNN Based CU Partition Mode decision algorithm for HEVC Inter Coding,” IEEE 2018) discloses the computer-implemented method of claim 1, wherein the RD cost is represented in a binary value, which indicates either the improved RD performance or no improvement in RD performance (HOMO or SPLIT, Section 2; HOMO is no-split, i.e., no improvement in RD; and SPLIT is the opposite).
Regarding Claim 5, Li (NPL: “CNN Based CU Partition Mode decision algorithm for HEVC Inter Coding,” IEEE 2018) discloses the computer-implemented method of claim 1, wherein Step b) is conducted using a multilayer perceptron model (the CNN has a multi-layer perceptron, Section 2.2).
Regarding Claim 6, Li (NPL: “CNN Based CU Partition Mode decision algorithm for HEVC Inter Coding,” IEEE 2018) discloses the computer-implemented method of claim 5, wherein a plurality of features of the block that is contacted as a feature vector is provided to the MLP model in Step b) as input (C5 is adopted to concatenate the Network I, II, and III; inputs are 38 matrices, which are the output feature maps of the previous sub-networks, Section 2.2.4).
Regarding Claim 8, Li (NPL: “CNN Based CU Partition Mode decision algorithm for HEVC Inter Coding,” IEEE 2018) discloses the computer-implemented method of claim 5, wherein the MLP model is trained using multiple datasets (ten video test sequences are selected as the source of training samples, Section 2.2) with different quantization parameters (QP is applied as an input, Section 2.2).
Regarding Claim 9, Li (NPL: “CNN Based CU Partition Mode decision algorithm for HEVC Inter Coding,” IEEE 2018) discloses the computer-implemented method of claim 6, wherein all feasible feature combinations of the plurality of features is explored to obtain predictive outcomes (inherent: his is what a neural network does: it establishes patterns/classification based on combinations of the input features), where the predictive outcomes are stored in a video encoder that is adapted to carry out the computer-implemented method (the algorithm is integrated into an HEVC encoder, Section 3).
Regarding Claim 10, Li (NPL: “CNN Based CU Partition Mode decision algorithm for HEVC Inter Coding,” IEEE 2018) discloses the computer-implemented method of claim 1, wherein the plurality of partition modes includes one or more of quadtree (QT) (splitting of a 2Nx2N block into four NxN blocks, Section 2; i.e., quadtree in standards vernacular).
Li does not disclose, but Taabane (NPL “Low Complexity Learning-Based QTMTT Partitioning Scheme for Inter Coding in VVC Encoder,” IEEE 2024) teaches horizontal binary tree (HBT), vertical binary tree (VBT), horizontal ternary tree (HTT) and vertical ternary tree (VTT) (available split modes, Table 1, are QT, vertical binary, horizontal binary, vertical ternary, and horizontal ternary).
One of ordinary skill in the art before the application was filed would have been motivated to follow the quadtree partitioning of Li with the multitype tree partitioning of Taabane because Huang teaches that multi-type tree partitioning provides enhanced video compression over quad-tree-only partitioning for high resolution video (Section 1 second-to-last paragraph), providing higher quality video at faster speed to consumers, improving experience.
Regarding Claim 13, Li (NPL: “CNN Based CU Partition Mode decision algorithm for HEVC Inter Coding,” IEEE 2018) discloses a non-transitory computer-readable memory recording medium having computer instructions recorded thereon, the computer instructions, when executed on one or more processors, causing the one or more processors to perform operations (platform is Huawei RH5885, Intel Xeon processor and RAM, Section 3) …. The remainder of Claim 13 is rejected on the grounds provided in Claim 1.
Regarding Claim 14, the claim is rejected on the grounds provided in Claim 13.
Claim(s) 3 is rejected under 35 U.S.C. 103 as being unpatentable over Li (NPL: “CNN Based CU Partition Mode decision algorithm for HEVC Inter Coding,” IEEE 2018) in view of Taabane (NPL “Low Complexity Learning-Based QTMTT Partitioning Scheme for Inter Coding in VVC Encoder,” IEEE 2024) and Wang (NPL “A probability-based zero-block early termination algorithm for QSHVC,” IEEE 2023).
Regarding Claim 3, Li (NPL: “CNN Based CU Partition Mode decision algorithm for HEVC Inter Coding,” IEEE 2018) discloses the computer-implemented method of claim 2, wherein the block is a coding unit (CU) (CU, Abstract).
Li does not disclose, but Taabane (NPL “Low Complexity Learning-Based QTMTT Partitioning Scheme for Inter Coding in VVC Encoder,” IEEE 2024) teaches the plurality of features of the block comprising … CU size features (CU size, the CU is evaluated by each eligible classifier, based on CU size, Section III.B, p. 141096 left column), partition features (each eligible classifier—note that size also indicates depth, Section III.B, p. 141096 left column), and context features (each eligible classifier, Section III.B, p. 141096 left column; constraints on CU partition modes … prevent certain modes after specific divisions, Section II.A, p. 141090 both columns).
Li does not disclose, but Wang (NPL “A probability-based zero-block early termination algorithm for QSHVC,” IEEE 2023) teaches features of the block comprising an AZB (all-zero block) feature (early terminate the mode and depth selection processes, we propose a probability-based zero-block early termination algorithm, Section III, p. 471, right column).
One of ordinary skill in the art before the application was filed would have been motivated to supplement partition decision of Li as modified by Taabane with the AZB-early termination because Wang teaches that all zero blocks indicate that the candidate coding mode and partition are the best one and that the CU is predicted accurately, and ignoring this potential coding improvement limits the potential improvements in coding speed (p. 470 right column – p. 471 left column).
Claim 7 is rejected under 35 U.S.C. 103 as being unpatentable over Li (NPL: “CNN Based CU Partition Mode decision algorithm for HEVC Inter Coding,” IEEE 2018) in view of Taabane (NPL “Low Complexity Learning-Based QTMTT Partitioning Scheme for Inter Coding in VVC Encoder,” IEEE 2024) and Lee (NPL “Fast Mode Decision Method of Multiple Weighted Bi-Predictions Using Lightweight Multilayer Perceptron in Versatile Video Coding,” MDPI 2023).
Regarding Claim 7, Li (NPL: “CNN Based CU Partition Mode decision algorithm for HEVC Inter Coding,” IEEE 2018) discloses the computer-implemented method of claim 5, wherein the MLP model comprises two (MLP 6 and 7, Fig. 1) densely connected layers (full connected MLP, Section 2.2).
Li does not disclose, but Lee (NPL “Fast Mode Decision Method of Multiple Weighted Bi-Predictions Using Lightweight Multilayer Perceptron in Versatile Video Coding,” MDPI 2023) teaches with their neuron counts set to 32 and 16 respectively (two hidden layers with 30 and 15 noes, p. 7).
One of ordinary skill in the art before the application was filed would have been motivated to develop the MLP layers of Li using the neuron counts of Lee because Lee teaches that doing so generates a lightweight, highly accurate, and minimally complex algorithm (Abstract), improving the efficiency of the codec.
Claim(s) 11-12 are rejected under 35 U.S.C. 103 as being unpatentable over Li (NPL: “CNN Based CU Partition Mode decision algorithm for HEVC Inter Coding,” IEEE 2018) in view of Taabane (NPL “Low Complexity Learning-Based QTMTT Partitioning Scheme for Inter Coding in VVC Encoder,” IEEE 2024) and Jia (US PG Publication 2015/0304683).
Regarding Claim 11, Li (NPL: “CNN Based CU Partition Mode decision algorithm for HEVC Inter Coding,” IEEE 2018) discloses the computer-implemented method of claim 1.
Li does not disclose, but Jia (US PG Publication 2015/0304683) teaches before Step c) (calculating the distortion and the code rate includes [0035]), steps of:
f) generating a residual signal of the block (corresponding to the residual [0035]);
g) detecting if the block is an AZB based on the residual signal (all-zero residual, non-all-zero residual [0035]);
h) in response to detecting that the block is not an AZB (non-all-zero residual [0035]), transforming and quantizing the residual signal (performing transformation, quantization [0035]), and entropy-coding transmission coefficients of the block (entropy encoding [0035]); and
i) in response to detecting that the block is an AZB (all-zero residual [0035]), skipping operations of transforming and quantizing the residual signal, and entropy-coding transmission coefficients of the block (calculating the distortion and the code rate corresponding to the all-zero residual by performing entropy encoding on the all-zero residual [0035]).
One of ordinary skill in the art before the application was filed would have been motivated to calculate rate-distortion in Li using the algorithm in Jia because the rate in Jia is the actual rate based on the partition, mode, and code amount, improving the accuracy and precision of the cost of Li, resulting in improved quality coding.
Regarding Claim 12, Li (NPL: “CNN Based CU Partition Mode decision algorithm for HEVC Inter Coding,” IEEE 2018) discloses the computer-implemented method of claim 11.
Li does not disclose, but Jia (US PG Publication 2015/0304683) teaches further comprises, before the step of detecting if the block is an AZB, a step of choosing either an intra-prediction model or an inter-prediction model for encoding the block and generating the residue signal (intra frame encoding step 101 is before rate-distortion step 102 [0030]-[0032]).
One of ordinary skill in the art before the application was filed would have been motivated to calculate rate-distortion in Li using the algorithm in Jia because the rate in Jia is the actual rate based on the partition, mode, and code amount, improving the accuracy and precision of the cost of Li, resulting in improved quality coding.
Response to Arguments
Applicant’s remarks filed 6/25/2026 have been considered.
Applicant’s remarks regarding Taabane and recursive quad-tree partitioning before switching to multi-type tree is persuasive, and another reference is relied upon instead. Remarks at 5-6. Applicant’s remarks do not pertain to the combination of references presented in this office action.
Applicant’s arguments regarding Wang are unpersuasive as they are not applicable to the combination of refences presented in this office action. Remark at 7. In addition, the argument that transplanting Wang’s AZB detection would produce inaccurate results in Taabane’s, or another framework, is unpersuasive; that Wang’s statistical assumptions do not hold is unpersuasive; Wang’s failure to detect PAZB is unpersuasive; Wang’s applicability to only square blocks is unpersuasive. Remarks at 7.
These arguments are predicated on arguing limitations that are not claimed, and the assumption that Wang must be bodily incorporated into the primary reference. Neither assumption is true. The combination does not need to be effective on non-square blocks because the claim does not include a limitation on being effective on non-square blocks. Likewise, the claim does not require detecting PAZB. Wang’s statistical assumptions are not required to be imported into the combination with the primary reference because bodily incorporation is not required.
A prima facie case of obviousness is based on what the references would have suggested. And Wang suggests that considering all-zero blocks to early-terminate a partition decision is effective because all-zero residual blocks tend to demonstrate excellent rate-distortion trade-off. That is all that the claim requires, and it is what Wang suggests. Therefore Wang supplies a motivation to consider all-zero blocks in partitioning, rendering the combination of references obvious against the language of Claim 3.
Applicant’s remaining arguments are not pertinent to the combination of references presented in this office action.
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure.
Huang (NPL “Learning -based fast splitting and directional mode decision for VVC Intra prediction,” IEEE 2024) – CU partitioning with neural network
Sun (NPL, “QTMT-LNN: A fast intra CU partition using lightweight neural network for 360-degree video coding on VVC,” IET Image Processing 2022) – CU partitioning with neural network.
THIS ACTION IS MADE FINAL. Applicant is reminded of the extension of time policy as set forth in 37 CFR 1.136(a).
A shortened statutory period for reply to this final action is set to expire THREE MONTHS from the mailing date of this action. In the event a first reply is filed within TWO MONTHS of the mailing date of this final action and the advisory action is not mailed until after the end of the THREE-MONTH shortened statutory period, then the shortened statutory period will expire on the date the advisory action is mailed, and any nonprovisional extension fee (37 CFR 1.17(a)) pursuant to 37 CFR 1.136(a) will be calculated from the mailing date of the advisory action. In no event, however, will the statutory period for reply expire later than SIX MONTHS from the mailing date of this final action.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to SHADAN E HAGHANI whose telephone number is (571)270-5631. The examiner can normally be reached M-F 9AM - 5PM.
Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at http://www.uspto.gov/interviewpractice.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Jay Patel can be reached at 571-272-2988. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of published or unpublished applications may be obtained from Patent Center. Unpublished application information in Patent Center is available to registered users. To file and manage patent submissions in Patent Center, visit: https://patentcenter.uspto.gov. Visit https://www.uspto.gov/patents/apply/patent-center for more information about Patent Center and https://www.uspto.gov/patents/docx for information about filing in DOCX format. For additional questions, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000.
/SHADAN E HAGHANI/Examiner, Art Unit 2485
Prosecution Insights