DETAILED ACTION
This action is responsive to the Amendments and Remarks received 12/30/2025 in which claims 11 and 14 are cancelled, claims 1, 3, 4, 6, 7, 9, 12, 13, and 15 are amended, and no claims are added as new claims.
Response to Arguments
On page 7 of the Remarks, Applicant contends neither Zhang nor Liu render obvious the features added by way of amendment to representative claim 1. Examiner disagrees. First, Applicant’s arguments are moot in view of the new grounds of rejection necessitated by amendment. The rejection now additionally relies on the teaches of Chujoh to teach or suggest the averred feature. See 35 U.S.C. 103 rejection, infra. Furthermore, in teaching that an index into a candidate weight set may be transmitted based on inter prediction mode or block size, Zhang is teaching that there is no need to initialize a candidate weight list for blocks that inherit the weight (e.g. in merge mode) from a neighboring block. Therefore, Zhang is not silent regarding how a weight candidate list is constructed as averred. Further still, Applicant does not argue that which is claimed. A list of values being constructed in the computer-related arts does not really have a “how” regarding its construction and Applicant’s claims and Specification are silent regarding any particular type of construction method as Applicant’s argument seems to suggest. The claim merely requires the candidate list be “configured based on” certain conditions, so there is no requirement regarding any particular type of construction of the list. Under a broadest reasonable interpretation whether the list is configured at all is covered by the claim language. And, in light of the teachings of the prior art, the most reasonable interpretation is that the list can be configured to support (1) only an equal weight, which is taught in the prior art for certain conditions; (2) a longer list [7 or 5] or shorter list [3], based on whether the pictures are low-delay pictures (low-delay might mean only previous picture as reference picture) and based on bi-prediction; or (3) no candidate list per se or a candidate list that is made up of weights that could be inherited from neighboring blocks using merge mode. For these reasons, Applicant’s argument that Zhang is deficient for failing to teach how a construction method of a candidate weight list differs under the recited conditions does not argue that which is claimed. Similarly, in arguing that Liu is deficient for failing to teach a mechanism for constructing the weight candidate list, Applicant does not argue that which is claimed. For all the foregoing reasons, the rejection under 35 U.S.C. 103 is reasonable and thus sustained.
Other claims are not argued separately. Remarks, 8.
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 1–7, 12, 13, and 15 are rejected under 35 U.S.C. 103 as being unpatentable over Zhang (US 2022/0239899 A1), Liu (US 2021/0203945 A1), and Chujoh (US 2022/0167002 A1).
Regarding claim 1, the combination of Zhang, Liu, and Chujoh teaches or suggests an image decoding method, the method comprising: generating a first prediction block of a current block based on a first prediction mode; generating a second prediction block of the current block based on a second prediction mode (Examiner interprets the recited first and second modes consistent with Applicant’s Specification and original claim 2 wherein the first and second predictions represent bi-prediction from List 0 and List 1 reference picture lists; Zhang, ¶¶ 0021 and 0115: teaches first and second reference samples based on two different motion vectors, each in different directions (List 0; List 1), which form a weighted prediction; see also Zhang, Fig. 8: illustrating bi-prediction in different directions); determining a weight for the current block based on a weight candidate list including a plurality of pre-defined weight candidates (Zhang, ¶ 0022: teaches the weights applied to the first and second predictions are determined according to BCW weight indexes; Examiner notes Applicant’s published paragraph [0105] explains the pre-defined weight candidates are prior art bcw_idx; Liu, ¶ 0133: teaches GBi, which is synonymous with BCW, or BWI, or GBIdx, etc., is used with DMVR, which is known to have constraints like size, inter-prediction mode, etc. and is used to generate a weighted average of templates; Liu, ¶¶ 0329, 0330, and 0337: teaches reordering the weighting factors in ascending order wherein the weighting factors are determined according to SAD and wherein the smallest SAD is selected as the weighting factor for the block); and generating a third prediction block of the current block by weighted summing of the first prediction block and the second prediction block based on the determined weight (Zhang, ¶ 0027: teaches first and second predictions can be combined using a weighted sum of template matching costs to form a final template matching cost for the bi-prediction candidate), wherein the weight candidate list is configured based on an inter prediction mode of the current block, whether a bi-prediction is applied to the current block, or a size of the current block (Zhang, ¶ 0239: teaches the size of the current block informs whether the reordering process is performed; Zhang, ¶¶ 0008–0009 and 0029: teaches size of the current block, inter-prediction mode, and whether bi-prediction (as opposed to uni-prediction) is indicated influence the use of the motion information in the manner described in Zhang; Liu, ¶ 0133: teaches GBi, which is synonymous with BCW, or BWI, or GBIdx, etc., is used with DMVR, which is known to have constraints like size, inter-prediction mode, etc. and is used to generate a weighted average of templates; Liu, ¶¶ 0329, 0330, and 0337: teaches reordering the weighting factors in ascending order wherein the weighting factors are determined according to SAD and wherein the smallest SAD is selected as the weighting factor for the block; While the combination of Zhang and Liu teach or suggest the weight candidate list configured based on inter-prediction mode, as broadly interpreted, Chujoh, ¶ 0291: more clearly establishes that GBi prediction can be applied when a CU is a certain size and further Chujoh, ¶¶ 0291 and 0299: teaches that bi-prediction mode differs from inter-prediction merge mode wherein inter-prediction merge mode inherits the weights from the merge candidate rather than referencing a GBi index whereas bi-prediction utilizes the signaled gbiIdx and the gbiWLut look up table).
One of ordinary skill in the art, before the effective filing date of the claimed invention, would have been motivated to combine the teachings of Zhang, with those of Liu, because both references are drawn to the same field of endeavor and because the skilled artisan would have been able to arrive at Applicant’s claimed reordered weight candidate list according to cost wherein the reordering is done in ascending order because while Zhang explains that reordering the candidate list according to template matching cost (Zhang, ¶ 0015) can take into account the BCW weight applied to the candidates when reordering the candidates in ascending order of costs (¶¶ 0212–0213), Liu specifically explains that the weighting factors themselves can be thought of as being reordered in ascending order to select the smallest SAD cost weighting factor. Thus, the combination of Zhang and Liu teaches reordering the BCW weighted candidate list just like Applicant’s published paragraph [0105] explains the claimed invention is one in which the weighted candidate list is reordered according to the ascending order and according to the same SAD cost metric. Therefore, the combination of Zhang and Liu represents a mere combination of prior art elements, according to known methods, to yield the predictable result of a set of candidates ordered based on the SAD cost metric (a similarity metric). This rationale applies to all combinations of Zhang and Liu 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 Zhang and Liu, with those of Chujoh, because all three references are drawn to the same field of endeavor such that one wishing to practice generalized bi-prediction with a candidate weight set would be led to their relevant teachings and because, as Chujoh explains, it was known that the coding mode or size of the block can inform how and whether the weights of GBi are applied. Thus, 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 Zhang, Liu, and Chojoh used in this Office Action unless otherwise noted.
Regarding claim 2, the combination of Zhang, Liu, and Chujoh teaches or suggests the method of claim 1, wherein: the first prediction mode represents a prediction in a L0 direction, and the second prediction mode represents a prediction in a L1 direction (Zhang, ¶¶ 0021 and 0115: teaches first and second reference samples based on two different motion vectors, each in different directions (List 0; List 1), which form a weighted prediction; see also Zhang, Fig. 8: illustrating bi-prediction in different directions).
Regarding claim 3, the combination of Zhang, Liu, and Chujoh teaches or suggests the method of claim 1, wherein: determining the weight includes calculating Sum of Absolute Difference (SDA) values corresponding to the plurality of weight candidates, respectively, the SAD values are calculated based on a reference template region and a template region of the current block, the reference template region is derived by weighted summing a template region of a L0 reference block and a template region of a L1 reference block based on a weight candidate (Zhang, ¶¶ 0024, 0026, 0027: teaches SAD between the current template and reference template for first and second prediction and calculating a weighted sum for a final template matching cost; see also Zhang, ¶ 0504: teaches the SAD cost is between the template and the reference template).
Regarding claim 4, the combination of Zhang, Liu, and Chujoh teaches or suggests the method of claim 3, wherein: a reordered weight candidate list is generated by reordering the plurality of weight candidates in an ascending order of the SAD values, the weight used for the current block is determined from the reordered weight candidate list (Zhang, ¶ 0213: teaches the template matching cost (SAD) for the candidates is reordered in an ascending order of costs).
Regarding claim 5, the combination of Zhang, Liu, and Chujoh teaches or suggests the method of claim 3, wherein: the method further includes obtaining a flag indicating whether a weighted prediction is performed for the current block from a bitstream (Zhang, ¶ 0010 and 0500: teaches differing weight indexes are only decoded if a flag indicates unequal weighting and that an embodiment contemplates only equal weighted averaging between L0 and L1 is available).
Regarding claim 6, the combination of Zhang, Liu, and Chujoh teaches or suggests the method of claim 5, wherein: when the flag indicates that the weighted prediction is performed for the current block, the weight used for the current block is determined based on a weight candidate having a smallest SAD value among the SAD values (Zhang, ¶ 0010 and 0500: teaches differing weight indexes are only decoded if a flag indicates unequal weighting and that an embodiment contemplates only equal weighted averaging between L0 and L1 is available; Zhang, ¶ 0213: teaches the template matching cost (SAD) for the candidates is reordered in an ascending order of costs; Zhang, ¶ 0214: explains the best candidate can be determined based on cost, which the skilled artisan knows the best candidate is the one having the smallest cost).
Regarding claim 7, the combination of Zhang, Liu, and Chujoh teaches or suggests the method of claim 5, wherein: when the flag indicates that the weighted prediction is not performed for the current block, a weight applied to the first prediction block and a weight applied to the second prediction block are determined as a same value (Zhang, ¶ 0010 and 0500: teaches differing weight indexes are only decoded if a flag indicates unequal weighting and that an embodiment contemplates only equal weighted averaging between L0 and L1 is available; Zhang, ¶ 0500: teaches the weights for bi-prediction may be equal weighted averaging).
Regarding claim 12, the combination of Zhang, Liu, and Chujoh teaches or suggests the method of claim 1, wherein: the current block is divided into a plurality of partitions based on a geometric partitioning mode, the plurality of partitions include a first partition and a second partition, the first prediction mode represents a prediction using prediction information of the first partition, and the second prediction mode represents a prediction using prediction information of the second partition (Zhang, ¶ 0008: teaches GPM).
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 15 lists the same elements as claim 1, but is drawn to a corresponding method to transforming a bitstream according to the coding method of claim 1. Therefore, the rationale for the rejection of claim 1 applies to the instant claim.
Claims 8–10 are rejected under 35 U.S.C. 103 as being unpatentable over Zhang, Liu, Chujoh, and Zhao (US 2022/0232233 A1).
Regarding claim 8, the combination of Zhang, Liu, Chujoh, and Zhao teaches or suggests the method of claim 3, wherein: the method further includes determining whether a weighted prediction is available for the current block based on a predefined condition (Zhao, ¶ 0140: teaches determining whether WP is enabled or not for the current block; Chujoh, ¶ 0291: teaches an SPS flag signaling whether GBi is possible and a CU level flag gbiAppliedFlag that can be set equal to TRUE or FALSE based on predefined conditions; While Chujoh teaches the feature under BRI, Zhao is cited because its teachings represent perhaps a different condition).
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 Zhang, Liu, and Chujoh, with those of Zhao, because all four references are drawn to the same field of endeavor and because, as Zhao explains, understanding whether weighted prediction (WP) or BCW is applied can be useful so that both techniques are not used simultaneously (Zhao, ¶ 0140). Thus, 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 Zhang, Liu, Chujoh, and Zhao used in this Office Action unless otherwise noted.
Regarding claim 9, the combination of Zhang, Liu, Chujoh, and Zhao teaches or suggests the method of claim 8, wherein: when the weighted prediction is available for the current block, the weight used for the current block is determined based on a weight candidate having a smallest SAD value among the SAD values (Zhao, ¶¶ 0020, 0165, and 0167: teaches the cost for weighted predictions is minimized to choose the best prediction; Liu, ¶ 0133: teaches GBi, which is synonymous with BCW, or BWI, or GBIdx, etc., is used with DMVR, which is known to have constraints like size, inter-prediction mode, etc. and is used to generate a weighted average of templates; Liu, ¶¶ 0329, 0330, and 0337: teaches reordering the weighting factors in ascending order wherein the weighting factors are determined according to SAD and wherein the smallest SAD is selected as the weighting factor for the block).
Regarding claim 10, the combination of Zhang, Liu, Chujoh, and Zhao teaches or suggests the method of claim 8, wherein: when the weighted prediction is not available for the current block, a weight applied to the first prediction block and a weight applied to the second prediction block are determined as a same value (Zhang, ¶ 0500: teaches the weights for bi-prediction may be equal weighted averaging; Zhao, ¶ 0126: teaches regular bi-prediction without explicit weighting parameters uses simple averaging, which is equal weights applied to the two prediction signals; see also Zhao, ¶ 0140: teaching that when BCW is not applied, equal weight can be inferred).
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure.
Hannuksela (US 2015/0304665 A1) teaches weighted prediction parameters can be signaled in a slice header, picture header, PPS, APS, or any other similar syntax structure (¶ 0455).
Bordes (US 2022/0060743 A1) teaches both weighted prediction and generalized bi prediction can be signaled using a flag at the PPS and that for weighted prediction, the flag is a weighted_pred_flag (¶ 0050 and Table 2).
Furht (US 2021/0168409 A1) teaches there are several different geometric patterns based on angle of segmenting line (Fig. 9) and teaches signaling indices for indicating an angle along a partition line segment dividing two partitions (¶0058).
Yu (US 2016/0255359 A1) teaches that the skilled artisan is aware of how to treat weighted prediction in the art such that the potentially larger bit depth resulting from weighted prediction needs to be rounded, right-shifted, and clipped back down to the original bit depth (¶ 0156).
Pu (US 2015/0098503 A1) teaches weighted prediction is based on bit depth of the input video and that right shifts are used to control overflow from multiplication operations (see e.g. ¶¶ 0019 and 0056).
Sato (US 2017/0034525 A1) teaches that granularity finer than the bit depth is desirable for weighted prediction (e.g. ¶ 0064).
Kim (US 2014/0140409 A1) teaches increasing bit depth for motion compensation and reducing the bit-depth back to picture bit-depth either before or after calculating the weighted average (¶ 0108).
Zhang (US 2022/0295088 A1) teaches bi-prediction template matching between the current block and a frame from reference list 0 and between the current frame and a frame from reference list 1 (¶ 0220). Zhang, ¶ 0156: teaches that a template matching candidate can be inserted at the beginning of the list and can also include a second candidate. Zhang, ¶¶ 0210 and 0300: teaches TM-DMVD is template matched decoder-side motion vector derivation which means it is not signaled but derived at the decoder side and can be used for both merge and AMVP modes of inter-prediction. Zhang, ¶ 0008: explains that template matching calculates a cost (difference) between the current block’s template and a reference template; Zhang, Fig. 23B: illustrates the template and reference template(s) are the same size; Zhang, ¶ 0156: explains the closest match (cost) between the current template and reference template is chosen. Zhang, ¶ 0112 et. seq. teaches geometric partitioning.
Chiang et al., “CE10.1.1: Multi-hypothesis prediction for improving AMVP mode, skip or merge mode, and intra mode,” JVET-L0100-v3, 12th Meeting: Macao, CN, Oct. 2018. Chiang teaches the final prediction is a weighted average of two merge candidates wherein the first inter-prediction merge candidate in the list is weighted heavier than the second inter-prediction merge candidate (Abstract and Section 2.2).
Chang (US 2022/0201282 A1) teaches equal and unequal weighted bi-prediction (¶ 0087).
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 extension fee 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 Michael J Hess whose telephone number is (571)270-7933. The examiner can normally be reached 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 571-273-8933.
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/Examiner, Art Unit 2481