DETAILED ACTION
Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Response to Arguments
Applicant amended claims 1, 3, 12 – 14, 25, 27, and 30 – 32 beyond formalities.
Applicant canceled claims 4 – 5, 7 – 9, 11, 15 – 19, and 22 – 24.
The pending claims are 1 – 3, 6, 10, 12 – 14, 20 – 21, and 25 – 34 [Page 9 lines 1 – 12].
Applicant amended the Drawings by adding two new figures to address Examiner’s Drawing Objections [Page 9 lines 13 – 14]. While the Drawings added merely restates the claim language with no other Specification support, the Examiner in the sole interest to expedite prosecution considers the Objection overcome.
Applicant does NOT comment on Examiner’s determination the claimed “processor” does NOT invoke Functional Analysis. In view of the apparent agreement, the Examiner in the sole interest of brevity deletes the Functional Analysis section.
Applicant’s arguments with respect to claim(s) 25 have been considered but are moot because the new ground of rejection does not rely on any reference applied in the prior rejection of record for any teaching or matter specifically challenged in the argument.
First, the Applicant recites the references against the claims [Page 9 lines 15 – 20].
Second, the Applicant contends amended claim 25 is not taught by the Mao and Misra combination [Page 9 lines 20 – 23], provides their interpretation of amended independent claim 25 [Page 9 line 24 – Page 10 line 7], and recites amended independent claim 25 [Page 10 lines 8 – 21].
Third, the Applicant broadly contends Mao lacks alleged features in the amended independent claims [Page 10 line 22 – Page 11 line 2]. The Examiner notes the features alleged not taught are not claimed and thus the argument is unpersuasive as a two level codec is taught (e.g. base and enhancement layer codecs such as in Figures 16 – 17 and associated description). However, in the sole interest to expedite prosecution, an additional reference is cited against the claims.
In response to applicant's arguments against the references individually, one cannot show nonobviousness by attacking references individually where the rejections are based on combinations of references. See In re Keller, 642 F.2d 413, 208 USPQ 871 (CCPA 1981); In re Merck & Co., 800 F.2d 1091, 231 USPQ 375 (Fed. Cir. 1986).
In response to applicant's argument that the references fail to show certain features of the invention, it is noted that the features upon which applicant relies (i.e., “opposite” allegations [Page 10 line 29 – Page 11 line 2]) are not recited in the rejected claim(s). Although the claims are interpreted in light of the specification, limitations from the specification are not read into the claims. See In re Van Geuns, 988 F.2d 1181, 26 USPQ2d 1057 (Fed. Cir. 1993).
Fourth, the Applicant contends Misra does not render obvious features of amended independent claim 25 [Page 11 lines 3 – 7] and provides alleged benefits of the invention citing Specification Paragraph 32 [Page 11 lines 7 – 4]. The Examiner notes Misra in at least Paragraphs 37 – 41 teach the output of data for machine consumption using multi-dimensional sets / scalable video coding techniques thus rendering obvious features of amended independent claim 25.
In response to applicant's argument that the references fail to show certain features of the invention, it is noted that the features upon which applicant relies (i.e., hybrid NN and “traditional” coding [Page 11 lines 3 – 6]) are not recited in the rejected claim(s). Although the claims are interpreted in light of the specification, limitations from the specification are not read into the claims. See In re Van Geuns, 988 F.2d 1181, 26 USPQ2d 1057 (Fed. Cir. 1993).
Fifth, the Applicant concludes the references do not render obvious the features of amended independent claim 25 [Page 11 lines 15 – 26] and contends the other amended independent claims are similarly allowable [Page 11 line 27 – Page 12 line 2].
Applicant's arguments fail to comply with 37 CFR 1.111(b) because they amount to a general allegation that the claims define a patentable invention without specifically pointing out how the language of the claims patentably distinguishes them from the references.
In response to applicant's arguments against the references individually, one cannot show nonobviousness by attacking references individually where the rejections are based on combinations of references. See In re Keller, 642 F.2d 413, 208 USPQ 871 (CCPA 1981); In re Merck & Co., 800 F.2d 1091, 231 USPQ 375 (Fed. Cir. 1986).
While the Applicant’s points may possibly be understood; the Examiner respectfully disagrees; however, in the sole interest to expedite prosecution cites a new reference against the amended claims.
Priority
Applicant’s claim for the benefit of a prior-filed application under 35 U.S.C. 119(e) or under 35 U.S.C. 120, 121, 365(c), or 386(c) is acknowledged (US Provisional Application 63/403,463 filed on September 2nd, 2022).
Information Disclosure Statement
The information disclosure statement (IDS) submitted on June 12th, 2026 was filed before the mailing date of the Final Rejection (this Office Action). The submission is in compliance with the provisions of 37 CFR 1.97. Accordingly, the information disclosure statement is being considered by the Examiner.
The information disclosure statement (IDS) submitted on February 21st, 2025 was filed before the mailing date of the First Action on the Merits (mailed May 18th, 2026). The submission is in compliance with the provisions of 37 CFR 1.97. Accordingly, the information disclosure statement is being considered by the Examiner.
Due to the excessively lengthy Information Disclosure Statement submitted by applicant, the examiner has given only a cursory review of the listed references. In accordance with MPEP 609.04(a), applicant is encouraged to provide a concise explanation of why the information is being submitted and how it is understood to be relevant. Concise explanations (especially those which point out the relevant pages and lines) are helpful to the Office, particularly where documents are lengthy and complex and applicant is aware of a section that is highly relevant to patentability or where a large number of documents are submitted and applicant is aware that one or more are highly relevant to patentability. Applicant is required to comply with this statement for any non-English language documents. See 37 CFR § 1.56 Duty to Disclose Information Material to Patentability.
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.
This application currently names joint inventors. In considering patentability of the claims the examiner presumes that the subject matter of the various claims was commonly owned as of the effective filing date of the claimed invention(s) absent any evidence to the contrary. Applicant is advised of the obligation under 37 CFR 1.56 to point out the inventor and effective filing dates of each claim that was not commonly owned as of the effective filing date of the later invention in order for the examiner to consider the applicability of 35 U.S.C. 102(b)(2)(C) for any potential 35 U.S.C. 102(a)(2) prior art against the later invention.
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 – 3, 6, 10, 12 – 14, 20 – 21, and 25 – 34 are rejected under 35 U.S.C. 103 as being unpatentable over Mao, et al. (US PG PUB 2025/0030879 A1 referred to as “Mao” throughout), and further in view of Misra, et al. (US PG PUB 2024/0380923 A1 referred to as “Misra” throughout) and Jang, et al. (US PG PUB 2025/0301145 A1 referred to as “Jang” throughout).
Note: Specification Paragraph 39 renders obvious variants to one of latent elements / tensors as being feature maps thus the teachings of latents or feature maps renders obvious the claimed “latent tensor” to one of ordinary skill in the art.
Regarding claim 1, see claim 25 which is the apparatus performing the steps of the claimed method.
Regarding claim 2, see claim 26 which is the apparatus performing the steps of the claimed method.
Regarding claim 3, see claim 27 which is the apparatus performing the steps of the claimed method.
Regarding claim 6, see claim 28 which is the apparatus performing the steps of the claimed method.
Regarding claim 10, see claim 29 which is the apparatus performing the steps of the claimed method.
Regarding claim 12, see claim 30 which is the apparatus performing the steps of the claimed method.
Regarding claim 13, see claim 31 which is the apparatus performing the steps of the claimed method.
Regarding claim 14, see claim 32 which is the apparatus performing the steps of the claimed method.
Regarding claim 20, see claim 33 which is the apparatus performing the steps of the claimed method.
Regarding claim 21, Mao teaches several configurations for encoding / decoding feature map / tensor data for human or machine usage depending on the intended user. Misra teaches the use of tensors to represent feature maps and prediction techniques for scalable encoding to modify Mao such as including inter-layer prediction techniques. Jang further renders obvious the teachings of Mao and Misra regarding the sampling and numbers of images / sub-images given as outputs in layers.
It would have been obvious to one of ordinary skill art before the effective filing date of the claimed invention to modify the architectures and neural network encoders / decoders of Mao with the tensor representations of Misra and prediction and sampling techniques additionally taught on the tensors / neural network output data and understanding using different numbers of images / outputs in each layer as taught by Jang. The combination teaches
a syntax element indicates that inter-layer prediction is only included when an image is available at a base layer [Misra Figures 5 and 8 – 13 (see reconstruction loops within the encoders including autoencoders as the NN based encoder before prediction) and 25 – 26 (encode / decode to reconstruct images and then further process for transmission) as well as Paragraphs 225 (syntax elements with conditions on signaling the use of inter layer prediction which relied on the base layer for reference in prediction described also in Paragraphs 41 and 181), and 233 – 238 (inter layer prediction between base / enhancement layer information based on reconstructed layers prior rendering obvious the first reconstructed version claimed and see inferred inter layer signaling between base / feature layers in Paragraph 236)].
See claim 12 for the motivation to combine Mao, Misra, and Jang.
Regarding claim 25, Mao teaches several configurations for encoding / decoding feature map / tensor data for human or machine usage depending on the intended user. Misra teaches the use of tensors to represent feature maps and prediction techniques for scalable encoding to modify Mao such as including inter-layer prediction techniques. Jang further renders obvious the teachings of Mao and Misra regarding the sampling and numbers of images / sub-images given as outputs in layers and provides another interpretation of the outputs suggested by at least Misra.
It would have been obvious to one of ordinary skill art before the effective filing date of the claimed invention to modify the architectures and neural network encoders / decoders of Mao with the tensor representations of Misra and prediction and sampling techniques additionally taught on the tensors / neural network output data and understanding using different numbers of images / outputs in each layer as taught by Jang with different uses of the outputs to further render obvious Misra’s teachings. The combination teaches
one or more processors [Mao Figures 2 – 4 (subfigures included see at least reference characters 20, 30, 43 and 44) as well as Paragraphs 171 – 179 (processor based encoder / decoder implementations)] and
at least one memory coupled to said one or more processors [Mao Figures 2 – 4 (subfigures included see at least reference characters 20, 30, 43, 44, 430, and 440) as well as Paragraphs 171 – 179 (processor based encoder / decoder implementations couple to or using memories)], wherein said one or more processors are configured to:
encode an image with a neural-network (NN) based method to generate a first output for a base layer [The Examiner notes the “for a base layer” is intended use language and thus carries no patentable weight; Mao Figures 2 – 3, 10 – 14 (subfigures included where Figures 10 – 11 at least are various neural network based encoders), and 16 – 18 as well as Paragraphs 181 – 185 and 218 – 220 (neural network implemented encoders / decoders to generate feature map outputs or decoding such encoded inputs), 276 – 280 and 287 – 290 (encoding an image with a neural network encoder for an enhancement layer to generate and output data), 306 – 312 and 329 – 336 (alternative architectures with Encoder2 used with same / similar functions using NN implementations such as in Figures 10 – 11 to generate output data) where the encoders may be implemented by CNNs as taught by Misra Paragraph 63 and Misra Paragraphs 40 – 42 (base layer for machine consumable data) and Jang Figure 2 at least where the base layer may be regarded as the bottom layer (NN encoder) as well as Paragraphs 65 – 68 (VCM encoder / video coding for machines to generate machine consumable data)], wherein said neural-network performs NN-based analysis [Mao Figures 8 – 14 (see at least NN encoders in Figures 10 – 11) as well as Paragraphs 184 – 185 (NN coding) and 217 – 222 (analysis / parameters with the feature map as part of NN coding); Misra Figures 6 – 7 (subfigures included) as well as Paragraphs 37,45, and 63 (analysis by NN autoencoder to generate base layer machine consumable data to incorporate into the structure of Jang Figure 2); Jang Figures 2 and 30 as well as Paragraphs 222 – 223 (analysis in VCM / NN based encoder)];
obtain a first reconstructed version of said image corresponding to said first output [Mao Figures 14 – 18 (see at least the xd , xc, or xc1 outputs based on the y (and tilde y encoder outputs too) and xc inputs (base / reference layers)) as well as Paragraphs 286 – 297 (reconstructing image data on the feature information and encoded image), 306 – 319 (various methods to obtain reconstructed image and relationships between layers), 372 – 380, 390 and 410 (decoder to reconstruct images from inputs / encoder outputs / quantized latents); Alternatively Misra Figures 5 and 8 – 13 (see reconstruction loops within the encoders including autoencoders as the NN based encoder before prediction) and 25 – 26 (encode / decode to reconstruct images and then further process for transmission) as well as Paragraphs 55 – 58 (reconstructed block used for prediction), 60 and 82 – 84 (reconstructed video data for prediction)];
predict a block of said image based on said first reconstructed version of said image to form a predicted block [Mao Figures 14 – 18 (see the encoders / autoencoders to combine / modify be teachings of Misra) as well as Paragraphs 132 – 135 (prediction in NN encoder implementations) and 202 – 204 (inter / intra layer prediction); Misra Figures 5 and 8 – 13 (see reconstruction loops within the encoders including autoencoders as the NN based encoder before prediction) and 25 – 26 (encode / decode to reconstruct images and then further process for transmission) as well as Paragraphs 57 – 60 (inter / intra prediction in encoders with reconstruction blocks), 79 – 82 and 87 – 88 (inter / intra prediction), 181, 225, and 234 – 236 (inter layer prediction between base / enhancement layer information based on reconstructed layers)]; and
encode said block based on said predicted block to generate a second output for an enhancement layer [The Examiner notes the “for a base layer” is intended use language and thus carries no patentable weight; Mao Figures 14 – 18 (see at least the encoders / autoencoders (such as reference character “AE” or encoders on the left side / modified by other layers) to combine / modify be teachings of Misra and Jang) as well as Paragraphs 63 – 67 and 181 – 184 (entropy encoding feature information or image data for each layer), 233 – 237 and 313 – 321 (entropy encoding quantized feature maps / image data / layer information), 338 – 345 and 362 – 372 (encoding feature information for machine processing such as in Paragraph 133); Misra Figures 5 and 8 – 13 (see at least reference characters 200, 222, 506, and 710 for entropy encoding) as well as Paragraphs 43 – 45 (PIPE and other entropy encoding techniques to use) and 86 – 91 (entropy encoding outputs / combing base and enhancement layer outputs); Jang Figure 2 at least where the enhancement layer may be regarded as the top layer (image/video encoder in at least reference character 211) as well as Paragraphs 93 – 100 (encoder for image / human vision output)].
The motivation to combine Misra with Mao is to combine features in the same / related field of invention of reconstruction of feature data utilizing compression techniques [Misra Paragraphs 1 and 4] in order to improve image quality [Misra Paragraphs 37 and 81 where the Examiner observes at least KSR Rationales (D) or (F) are also applicable].
The motivation to combine Jang with Mao and Misra is to combine features in the same / related field of invention if feature encoding / decoding for machine task performance and human vision [Jang Paragraphs 1 – 2] in order to improve performance of image processing for machine only tasking [Jang Paragraphs 2 and 30 – 38 where the Examiner observes KSR Rationales (D) or (F) are applicable].
This is the motivation to combine Mao, Misra, and Jang which will be used throughout the Rejection.
Regarding claim 26, Mao teaches several configurations for encoding / decoding feature map / tensor data for human or machine usage depending on the intended user. Misra teaches the use of tensors to represent feature maps and prediction techniques for scalable encoding to modify Mao such as including inter-layer prediction techniques. Jang further renders obvious the teachings of Mao and Misra regarding the sampling and numbers of images / sub-images given as outputs in layers and provides another interpretation of the outputs suggested by at least Misra.
It would have been obvious to one of ordinary skill art before the effective filing date of the claimed invention to modify the architectures and neural network encoders / decoders of Mao with the tensor representations of Misra and prediction and sampling techniques additionally taught on the tensors / neural network output data and understanding using different numbers of images / outputs in each layer as taught by Jang with different uses of the outputs to further render obvious Misra’s teachings. The combination teaches
obtain a latent tensor corresponding to said image based on said neural-network based method [Mao Figures 2 – 3 and 14 – 18 (subfigures included additionally where Figures 10 – 11 at least are various neural network based encoders) as well as Paragraphs 181 – 185 and 218 – 220 (neural network implemented encoders / decoders to generate feature map outputs or decoding such encoded inputs), 276 – 280 and 287 – 290 (encoding an image with a neural network encoder for an enhancement layer to generate and output data), 306 – 312 and 329 – 336 (alternative architectures with Encoder2 used with same / similar functions using NN implementations such as in Figures 10 – 11 to generate output data) where the encoders may be implemented by CNNs as taught by Misra Figures 5 and 8 – 13 as well as Paragraph 63, 86 – 90 (outputting feature maps from encoding / decoding), 131 and 234 (feature tensors and feature maps obvious variants of the claimed latent tensor to one of ordinary skill in the art)];
quantize said latent tensor [Mao Figures 14 – 18 (see the “Q” blocks for quantization) as well as Paragraphs 288 – 297 (quantizing feature values / maps or residuals / predicted tensors / feature maps) and 359 – 362 (quantizing residuals / predictions of feature maps); Misra Figures 8 – 13 (see at least reference characters 502) as well as Paragraphs 86 – 90 (quantization of feature map done by the encoder / autoencoder)]; and
entropy code said quantized latent tensor to generate said first output [Misra Figures 14 – 18 (see at least the encoders / autoencoders (such as reference character “AE” or encoders on the left side / modified by other layers) coding output / feature maps as output of quantization steps and to combine / modify be teachings of Misra) as well as Paragraphs 63 – 67 and 181 – 184 (entropy encoding feature maps or image data for each layer), 233 – 237 and 313 – 321 (entropy encoding quantized feature maps / image data / layer information), 338 – 345 and 362 – 372 (encoding feature information for machine processing such as in Paragraph 133); Misra Figures 5 and 8 – 13 (see at least reference characters 200, 222, 506, and 710 for entropy encoding) as well as Paragraphs 43 – 45 (PIPE and other entropy encoding techniques to use) and 86 – 91 (entropy encoding outputs / combing base and enhancement layer outputs)].
See claim 25 for the motivation to combine Mao, Misra, and Jang.
Regarding claim 27, Mao teaches several configurations for encoding / decoding feature map / tensor data for human or machine usage depending on the intended user. Misra teaches the use of tensors to represent feature maps and prediction techniques for scalable encoding to modify Mao such as including inter-layer prediction techniques. Jang further renders obvious the teachings of Mao and Misra regarding the sampling and numbers of images / sub-images given as outputs in layers and provides another interpretation of the outputs suggested by at least Misra.
It would have been obvious to one of ordinary skill art before the effective filing date of the claimed invention to modify the architectures and neural network encoders / decoders of Mao with the tensor representations of Misra and prediction and sampling techniques additionally taught on the tensors / neural network output data and understanding using different numbers of images / outputs in each layer as taught by Jang with different uses of the outputs to further render obvious Misra’s teachings. The combination teaches
obtain a first reconstructed version of said image by performing a second neural-network based method on said quantized latent tensor, wherein said second neural-network based method synthesizes an image predictor from said quantized latent tensor [Mao Figures 13 – 18 (subfigures included – see the NN architecture of Figure 13 reconstructing based on the quantized feature map / latents and the decoder 1 and decoder2 reference characters in which in view of Paragraphs 148 (decoding network) and 127 – 138 (various NN techniques including DNN / CNN / RNN to form predictors from latents / feature maps further explained in Paragraphs 139 – 144) render obvious the decoding as a NN implementation) as well as Paragraphs 194 – 201 and 265 – 270 (correct / decoding network used to form reconstructed image based on residual feature maps), 272 – 278 (decoding network architectures); Misra Figures 7 – 12 (subfigures included see at east reference characters 412, 504, and 606 for the autodecoders) as well as Paragraphs 63 (forming predictors with MDDS / quantized latents), 80 and 86 – 90 (autodecoders use kernels / perform as NNs and form prediction data from MDDS data / quantized latents) and 239 (NN implementations including CNNs in Paragraph 63); Jang Figures 2 – 4 and 30 as well as Paragraphs 68 and 75 – 84 (synthesizing in VCM / NN encoders to combine with the image / video encoding / decoding in the enhancement layer in Figure 3 as well as Paragraphs 101 – 109 (forming predictors including using feature maps / latents))].
See claim 25 for the motivation to combine Mao, Misra, and Jang.
Regarding claim 28, Mao teaches several configurations for encoding / decoding feature map / tensor data for human or machine usage depending on the intended user. Misra teaches the use of tensors to represent feature maps and prediction techniques for scalable encoding to modify Mao such as including inter-layer prediction techniques. Jang further renders obvious the teachings of Mao and Misra regarding the sampling and numbers of images / sub-images given as outputs in layers and provides another interpretation of the outputs suggested by at least Misra.
It would have been obvious to one of ordinary skill art before the effective filing date of the claimed invention to modify the architectures and neural network encoders / decoders of Mao with the tensor representations of Misra and prediction and sampling techniques additionally taught on the tensors / neural network output data and understanding using different numbers of images / outputs in each layer as taught by Jang with different uses of the outputs to further render obvious Misra’s teachings. The combination teaches
select a prediction mode for said block, from intra prediction, temporal prediction and interlayer prediction [Mao Figures 14 – 18 (see the encoders / autoencoders to combine / modify be teachings of Misra) as well as Paragraphs 132 – 135 (prediction in NN encoder implementations), 142 (Intra frame prediction / intra prediction for selection with inter prediction) and 202 – 204 (inter / intra layer prediction); Misra Figures 5 and 8 – 13 (see reconstruction loops within the encoders including autoencoders as the NN based encoder before prediction and see selection between inter / intra prediction as in at least Figures 5 and 10) and 25 – 26 (encode / decode to reconstruct images and then further process for transmission) as well as Paragraphs 57 – 60 (inter / intra prediction in encoders with reconstruction blocks), 79 – 82 and 87 – 88 (inter / intra prediction), 181, 225, and 234 – 236 (inter layer prediction between base / enhancement layer information based on reconstructed layers)], wherein said predicting a block based on said first reconstructed version of said image is performed responsive to inter-layer prediction being selected [Misra Figures 5 and 8 – 13 (see reconstruction loops within the encoders including autoencoders as the NN based encoder before prediction) and 25 – 26 (encode / decode to reconstruct images and then further process for transmission) as well as Paragraphs 181, 225, and 233 – 238 (inter layer prediction between base / enhancement layer information based on reconstructed layers prior rendering obvious the first reconstructed version claimed)].
See claim 25 for the motivation to combine Mao, Misra, and Jang.
Regarding claim 29, Mao teaches several configurations for encoding / decoding feature map / tensor data for human or machine usage depending on the intended user. Misra teaches the use of tensors to represent feature maps and prediction techniques for scalable encoding to modify Mao such as including inter-layer prediction techniques. Jang further renders obvious the teachings of Mao and Misra regarding the sampling and numbers of images / sub-images given as outputs in layers and provides another interpretation of the outputs suggested by at least Misra.
It would have been obvious to one of ordinary skill art before the effective filing date of the claimed invention to modify the architectures and neural network encoders / decoders of Mao with the tensor representations of Misra and prediction and sampling techniques additionally taught on the tensors / neural network output data and understanding using different numbers of images / outputs in each layer as taught by Jang with different uses of the outputs to further render obvious Misra’s teachings. The combination teaches
said image is from a video sequence, and wherein said first output is generated for a first subset of images of said video sequence [Mao Figures 8 and 14 – 18 as well as Paragraphs 203 – 206 (images of video sequence as inputs); Misra Figures 16 and 23 as well as Paragraphs 166 (group of pictures from a video sequence as input)], and
said second output is generated for a second subset of images of said video sequence, wherein said second subset of images includes more images than said first subset of images [Mao Figures 16 and 23 as well as Paragraphs 139 – 143 (frame rates different for enhancement layers and the base layer rendering obvious the “more images” in scalable coding as would be readily recognized by one of ordinary skill in the art in combination with suggestions and figures of Jang); Misra Figures 16 and 23 as well as Paragraphs 166 (groups of pictures for coding / processing in each layer), and 228 – 238 (parameters including subsets of feature maps / tensors / picture to use between base / enhancement layers signaled); Jang Figures 17, 20, 21 – 24, and 25 as well as Paragraphs 182 – 186 and 202 – 204 (stem / layers of the processing for different layers of the encoder / decoder in which the stems are for the various enhancement layers of Mao / Misra as well)].
See claim 25 for the motivation to combine Mao, Misra, and Jang.
Regarding claim 30, Mao teaches several configurations for encoding / decoding feature map / tensor data for human or machine usage depending on the intended user. Misra teaches the use of tensors to represent feature maps and prediction techniques for scalable encoding to modify Mao such as including inter-layer prediction techniques. Jang further renders obvious the teachings of Mao and Misra regarding the sampling and numbers of images / sub-images given as outputs in layers and provides another interpretation of the outputs suggested by at least Misra.
It would have been obvious to one of ordinary skill art before the effective filing date of the claimed invention to modify the architectures and neural network encoders / decoders of Mao with the tensor representations of Misra and prediction and sampling techniques additionally taught on the tensors / neural network output data and understanding using different numbers of images / outputs in each layer as taught by Jang with different uses of the outputs to further render obvious Misra’s teachings. The combination teaches
one or more processors [Mao Figures 2 – 4 (subfigures included see at least reference characters 20, 30, 43 and 44) as well as Paragraphs 171 – 179 (processor based encoder / decoder implementations)] and
at least one memory coupled to said one or more processors [Mao Figures 2 – 4 (subfigures included see at least reference characters 20, 30, 43, 44, 430, and 440) as well as Paragraphs 171 – 179 (processor based encoder / decoder implementations couple to or using memories)], wherein said one or more processors are configured to:
entropy decode a latent tensor corresponding an image [Mao Figures 12 – 18 (see Figures 12 – 13 for NN based methods for decoding as well as the “AD” reference character (entropy decoding) processing the y / z labeled feature map / latent tensor inputs and “Decoder1” / “Decoder2” which may include use of decoders taught by Misra) as well as Paragraphs 146 – 150 (entropy decoding with NN implementations), 266 – 267 (autodecoding / entropy decoding), 295, 312, 335, 362, and 397 (entropy decoding in each layer); Misra Figures 4, 6, 8, and 11 – 14 (see at least reference characters 302, 602, and 802 for entropy decoders) as well as Paragraphs 87 – 92 (entropy decoding as first step in decoding process of feature maps / latents)];
obtain a first reconstructed version of said image for a base layer based on said latent tensor, using a neural network based method, wherein said neural-network based method performs synthesis [The Examiner notes the “for a base layer” is intended use language and thus carries no patentable weight; Mao Figures 14 – 18 (see at least the xd , xc, or xc1 outputs based on the y (and tilde y encoder outputs too) and xc inputs (base / reference layers)) as well as Paragraphs 126 – 131 (NN implementations of encoders / decoders), 286 – 297 (reconstructing image data on the feature information and encoded image), 306 – 319 (various methods to obtain reconstructed image and relationships between layers), 372 – 380, 390 and 410 (decoder to reconstruct images from inputs / encoder outputs / quantized latents); Alternatively Misra Figures 5 and 8 – 13 (see reconstruction loops within the decoders including autodecoders as the NN based prediction) and 25 – 26 (encode / decode to reconstruct images and then further process for transmission) as well as Paragraphs 55 – 58 (reconstructed block used for prediction), 60 and 82 – 87 (reconstructed video data for prediction and NN based autodecoders which also may be in encoding loops as well using NN based methods in Paragraphs 63 and 239 (e.g. CNN) combinable with Jang’s NN based decoders / VCM decoders); Jang Figures 2 – 4 (See at least reference character 223)) and 30 as well as Paragraphs 68 and 75 – 84 (synthesizing in VCM / NN encoders to combine with the image / video encoding / decoding in the base layer in Figure 3 as well as Paragraphs 101 – 109 (forming predictors including using feature maps / latents))];
predict a block of said image based on said first reconstructed version of said image to form a predicted block [Mao Figures 14 – 18 (see the encoders / autoencoders to combine / modify be teachings of Misra) as well as Paragraphs 132 – 135 (prediction in NN encoder implementations) and 202 – 204 (inter / intra layer prediction); Misra Figures 5 and 8 – 13 (see reconstruction loops within the encoders including autoencoders as the NN based encoder before prediction) and 25 – 26 (encode / decode to reconstruct images and then further process for transmission) as well as Paragraphs 57 – 60 (inter / intra prediction in encoders with reconstruction blocks), 79 – 82 and 87 – 88 (inter / intra prediction), 181, 225, and 234 – 236 (inter layer prediction between base / enhancement layer information based on reconstructed layers); Jang Figures 2 – 4 (See at least reference character 223)) and 30 as well as Paragraphs 68 and 75 – 84 (synthesizing in VCM / NN encoders to combine with the image / video encoding / decoding in the base layer in Figure 3 as well as Paragraphs 101 – 109 (forming predictors including using feature maps / latents))]; and
decode said block based on said predicted block to obtain a second reconstructed version of said image for an enhancement layer [The Examiner notes the “for a base layer” is intended use language and thus carries no patentable weight; Mao Figures 2 (see at least reference character 30), 6, and 13 – 18 (subfigures included – see at least “decoder1” and “decoder2” forming the reconstructed images from the feature maps entropy decoded) as well as Paragraphs 126 – 131 (NN implementations of encoders / decoders), 147 – 160 (decoders to undo AI encoding using AI techniques and full system implementation), 286 – 297 (reconstructing image data on the feature information and encoded image), 306 – 319 (various methods to obtain reconstructed image and relationships between layers), 339 – 348 (decoding a base / enhancement layer), 372 – 380 (“decoder2” in processing base / enhancement layer), 390 and 410 (decoder to reconstruct images from inputs / encoder outputs / quantized latents); Misra Figures 5 and 8 – 13 (see reconstruction loops within the decoders including autodecoders and at least reference characters 300 and 412A, and 412B) and 25 – 26 (encode / decode to reconstruct images and then further process for transmission) as well as Paragraphs 55 – 58 (decoding / reconstructing blocks for additional processing), 60 and 82 – 87 (reconstructed video data for prediction and NN based autodecoders which are in decoding loops using NN based methods in Paragraphs 63 and 239 (e.g. CNN)); Jang Figures 2 – 4 (see at least reference character 212) at least where the enhancement layer may be regarded as the top layer (image/video encoder in at least reference character 211) as well as Paragraphs 93 – 100 (encoder for image / human vision output)].
See claim 25 for the motivation to combine Mao, Misra, and Jang as similar motivation exists since decoding is the inverse process of encoding to one of ordinary skill in the art.
Regarding claim 31, Mao teaches several configurations for encoding / decoding feature map / tensor data for human or machine usage depending on the intended user. Misra teaches the use of tensors to represent feature maps and prediction techniques for scalable encoding to modify Mao such as including inter-layer prediction techniques. Jang further renders obvious the teachings of Mao and Misra regarding the sampling and numbers of images / sub-images given as outputs in layers and provides another interpretation of the outputs suggested by at least Misra.
It would have been obvious to one of ordinary skill art before the effective filing date of the claimed invention to modify the architectures and neural network encoders / decoders of Mao with the tensor representations of Misra and prediction and sampling techniques additionally taught on the tensors / neural network output data and understanding using different numbers of images / outputs in each layer as taught by Jang with different uses of the outputs to further render obvious Misra’s teachings. The combination teaches
wherein obtaining a first reconstructed version of said image comprises performing said neural-network based method based on said latent tensor [Mao Figures 13 – 18 (subfigures included – see the NN architecture of Figure 13 reconstructing based on the quantized feature map / latents and the “AD”, “decoder1” and “decoder2” reference characters in which in view of Paragraphs 148 – 159 (decoding network with autodecoder and NN based decoder implementations) and 127 – 138 (various NN techniques including DNN / CNN / RNN) render obvious the decoding as a NN implementation), 194 – 201 and 265 – 270 (correct / decoding network used to form reconstructed image based on residual feature maps), 272 – 278 (decoding network architectures); Misra Figures 7 – 12 (subfigures included see at east reference characters 412, 504, and 606 for the autodecoders) as well as Paragraphs 63, 80 and 86 – 90 (autodecoders use kernels / perform as NNs) and 239 (NN implementations including CNNs in Paragraph 63)].
See claim 30 for the motivation to combine Mao, Misra, and Jang.
Regarding claim 32, Mao teaches several configurations for encoding / decoding feature map / tensor data for human or machine usage depending on the intended user. Misra teaches the use of tensors to represent feature maps and prediction techniques for scalable encoding to modify Mao such as including inter-layer prediction techniques. Jang further renders obvious the teachings of Mao and Misra regarding the sampling and numbers of images / sub-images given as outputs in layers and provides another interpretation of the outputs suggested by at least Misra.
It would have been obvious to one of ordinary skill art before the effective filing date of the claimed invention to modify the architectures and neural network encoders / decoders of Mao with the tensor representations of Misra and prediction and sampling techniques additionally taught on the tensors / neural network output data and understanding using different numbers of images / outputs in each layer as taught by Jang with different uses of the outputs to further render obvious Misra’s teachings. The combination teaches
upscale output from said neural-network based method to obtain said first reconstructed version of said image [Mao Figures 8 – 13 (method and NNs with down and upscaling / up sampling) as well as Paragraphs 220 – 226 (up-sampling / resolution conversion as part of the final processing of decoding where upscaling / up-sampling are obvious variants to one of ordinary skill in the art); Misra Figures 16 and 22 – 23 as well as Paragraphs 132 and 165 – 166 (decoding with up-samplers as part of reconstructing the image / output for display)].
See claim 30 for the motivation to combine Mao, Misra, and Jang.
Regarding claim 33, Mao teaches several configurations for encoding / decoding feature map / tensor data for human or machine usage depending on the intended user. Misra teaches the use of tensors to represent feature maps and prediction techniques for scalable encoding to modify Mao such as including inter-layer prediction techniques. Jang further renders obvious the teachings of Mao and Misra regarding the sampling and numbers of images / sub-images given as outputs in layers and provides another interpretation of the outputs suggested by at least Misra.
It would have been obvious to one of ordinary skill art before the effective filing date of the claimed invention to modify the architectures and neural network encoders / decoders of Mao with the tensor representations of Misra and prediction and sampling techniques additionally taught on the tensors / neural network output data and understanding using different numbers of images / outputs in each layer as taught by Jang with different uses of the outputs to further render obvious Misra’s teachings. The combination teaches
said image is from a video sequence, and wherein said latent tensor is decoded for a first subset of images of said video sequence [Mao Figures 8 and 14 – 18 as well as Paragraphs 203 – 206 (images of video sequence as inputs); Misra Figures 16 and 23 as well as Paragraphs 166 (group of pictures from a video sequence as input)], and
said second reconstructed version of said image is performed for a second subset of images of said video sequence, wherein said second subset of images includes more images than said first subset of images [See claim 25 for citations of “second reconstructed version of said image” for citations and additionally Mao Figures 16 and 23 as well as Paragraphs 139 – 143 (frame rates different for enhancement layers and the base layer rendering obvious the “more images” in scalable coding as would be readily recognized by one of ordinary skill in the art in combination with suggestions and figures of Jang); Misra Figures 16 and 23 as well as Paragraphs 166 (groups of pictures for coding / processing in each layer), and 228 – 238 (parameters including subsets of feature maps / tensors / picture to use between base / enhancement layers signaled); Jang Figures 17, 20, 21 – 24, and 25 as well as Paragraphs 182 – 186 and 202 – 204 (stem / layers of the processing for different layers of the encoder / decoder in which the stems are for the various enhancement layers of Mao / Misra as well)].
See claim 30 for the motivation to combine Mao, Misra, and Jang.
Regarding claim 34, Mao teaches several configurations for encoding / decoding feature map / tensor data for human or machine usage depending on the intended user. Misra teaches the use of tensors to represent feature maps and prediction techniques for scalable encoding to modify Mao such as including inter-layer prediction techniques. Jang further renders obvious the teachings of Mao and Misra regarding the sampling and numbers of images / sub-images given as outputs in layers and provides another interpretation of the outputs suggested by at least Misra.
It would have been obvious to one of ordinary skill art before the effective filing date of the claimed invention to modify the architectures and neural network encoders / decoders of Mao with the tensor representations of Misra and prediction and sampling techniques additionally taught on the tensors / neural network output data and understanding using different numbers of images / outputs in each layer as taught by Jang with different uses of the outputs to further render obvious Misra’s teachings. The combination teaches
a syntax element indicates that interlayer prediction is only included when an image is available at a base layer [Misra Figures 5 and 8 – 13 (see reconstruction loops within the encoders including autoencoders as the NN based encoder before prediction) and 25 – 26 (encode / decode to reconstruct images and then further process for transmission) as well as Paragraphs 225 (syntax elements with conditions on signaling the use of inter layer prediction which relied on the base layer for reference in prediction described also in Paragraphs 41 and 181), and 233 – 238 (inter layer prediction between base / enhancement layer information based on reconstructed layers prior rendering obvious the first reconstructed version claimed and see inferred inter layer signaling between base / feature layers in Paragraph 236)].
See claim 30 for the motivation to combine Mao, Misra, and Jang.
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure. Hsiao (WO2024/049441 A1 referred to as “Hsiao” throughout) [First Cited in the Office Action mailed May 18th, 2026] teaches similar features to the pending application, but as the same effective filing date in Figure 2.
References considered which may raise ODP Rejections based on amendments made to the claims: Racape, et al. (US Patent #12,556,720 B2 referred to as “Racape” throughout) and Galpin, et al. (US Patent #11,973,964 B2 referred to as “Galpin” throughout).
Applicant's amendment necessitated the new ground(s) of rejection presented in this Office action. Accordingly, THIS ACTION IS MADE FINAL. See MPEP § 706.07(a). 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 Tyler W Sullivan whose telephone number is (571)270-5684. The examiner can normally be reached IFP.
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, David Czekaj can be reached at (571)-272-7327. 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.
/TYLER W. SULLIVAN/Primary Examiner, Art Unit 2487
Prosecution Insights