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 .
Priority
Receipt is acknowledged of certified copies of papers required by 37 CFR 1.55.
Response to Arguments
Applicant's arguments filed August 4, 2026 have been fully considered.
This final action is in response to the amendment filed August 4, 2026. Claims 1-18 are pending. Claims 1 and 10 are amended.
Regarding the § 101, applicant argues that data gathering is not an enumerated grouping and that generating scene images cannot be done in the mind. The comparison and selection steps are the exception. The obtaining steps are additional elements weighed at Prong Two and Step 2B.
Applicant argues that the claims improve positioning technology without satellites. The claim does not change the camera, the model that renders the second images, or the navigation application. The improvement identified is the screening procedure, which is the abstract idea itself. An improvement to the abstract idea is not an improvement to technology under MPEP 2106.05(a). In Desjardins the claims changed how the model was trained; here the model is used as constructed. The Recentive discussion is applied to claims 7 and 16, which recite the trained model.
Applicant argues that the Step 2B finding rested only on the applied references. The Step 2B findings now rest on express statements in the specification and on publications. The specification states that satellites are needed for navigation: “Currently, satellites are needed in navigation to perform perspective positioning” (SPEC ¶ 3). Maggio states that neural radiance fields encode 3D geometry and appearance (“Neural Radiance Fields (NeRF) have gained significant popularity, as they can encode both 3D geometry and appearance of an environment” (Maggio p. 1)), and Lin states that neural scene representations synthesize photo-realistic views (Lin p. 2).
Applicant’s arguments with respect to claim(s) 1-18 regarding Bao have been considered but are moot because the new ground of rejection does not rely on the Bao reference alone as applied in the prior rejection of record for teachings or matters specifically challenged in the argument.
Claim Rejections - 35 USC § 101
35 U.S.C. 101 reads as follows:
Whoever invents or discovers any new and useful process, machine, manufacture, or composition of matter, or any new and useful improvement thereof, may obtain a patent therefor, subject to the conditions and requirements of this title.
Claims 1-18 rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more. The claim(s) recite(s) data gathering, data assignment, mental processes, and mathematical concepts. This judicial exception is not integrated into a practical application because there is no particular machine, transformation, or specific technical mechanism is recited. The claim(s) does/do not include additional elements that are sufficient to amount to significantly more than the judicial exception because the specification and the publications cited below establish that the additional elements, individually and in combination, are well-understood, routine, and conventional.
Step 1
Claim 1 recites a process and falls within a statutory category. Claim 10 recites the same operations performed by a machine which also falls within a statutory category.
Step 2A, Prong 1
The claim recites an abstract idea in the groupings of mental processes and mathematical concepts:
“obtaining a first image collected from a target scene” – data gathering.
“obtaining a plurality of second images, each second image being an image corresponding to the target scene and not obtained by an image acquisition device for
the target scene” – data gathering; source of the images does not change what is done with them.
“determining a second target image from the plurality of second images according to the first image” – mental process (image comparison and selection) and mathematical concept (similarity computation).
“obtaining a plurality of third images, the plurality of third images being images corresponding to the target scene, and each third image meeting a first viewing angle condition with the second target image, wherein viewing angle data of each third image and viewing angle data of the second target image satisfy a viewing angle similarity condition” – data gathering constrained by a relational criterion; the similarity condition is a numeric comparison (SPEC ¶ 37: “a difference between the viewing angle data is less than or equal to a corresponding difference threshold”).
“determining a third target image from the plurality of third images according to the first image” – mental process and mathematical concept.
“determining viewing angle data corresponding to the first image according to viewing angle data of the third target image” – data assignment.
“outputting the determined viewing angle data corresponding to the first image as a
viewing angle positioning result in a navigation application” – post-solution output of
the assigned data.
The claim compares a query image to candidates, picks the closest match, narrows the search space, compares again, and copies the matched candidate's viewing-angle metadata to the query. A person can perform this in the mind, aided by ordinary similarity computation.
Step 2A, Prong 2
The additional elements do not integrate the exception into a practical application. No particular machine, transformation, or specific technical mechanism is recited. Obtaining the first image from a camera and obtaining second images that are not obtained by an image acquisition device are data gathering (MPEP 2106.05(g)); the claim does not change the camera or the model that produces the second images. Outputting the result as a positioning result in a navigation application is post-solution output and a field of use (MPEP 2106.05(g), (h)). The asserted improvement – substituting image matching for satellite-based positioning (Spec. ¶¶ 3-4) – is the abstract idea itself, not a technological improvement to a computer or another technology. Limiting the steps to “scene images” is a field-of-use restriction. The claim recites no similarity measure, no rendering technique and no navigation operation; it recites that images are compared, one is selected, and its recorded pose is assigned and output, and the specification describes the improvement as that screening itself (SPEC ¶ 99: “positioning is achieved based on the iteration method of image comparison”), unlike the specific animation rules of McRO or the sensor arrangement of Thales Visionix. As to claims 7 and 16, which recite a trained first model used as constructed (Spec. ¶ 42), See Recentive Analytics, Inc. v. Fox Corp., 134 F.4th 1205 (Fed. Cir. 2025); no improvement to the model itself is claimed, cf. Ex parte Desjardins, Appeal No. 2024-000567 (Appeals Review Panel Sept. 26, 2025). Ineligibility is established by a preponderance of the evidence.
Step 2B
The additional elements, individually and in combination, are well-understood, routine,
and conventional. The specification and the following publications establish this (MPEP 2106.05(d)(I)):
Collecting the query image with a camera (SPEC ¶ 21: “the first image is collected in a
shopping mall by an image acquisition device such as a camera”) (Bao ¶ 3: “surrounding rich visual information can be acquired by using a mobile phone and other devices for photographing images”).
Rendering candidate views of the scene from a model built for it rather than collecting them (SPEC ¶ 24: “The NERF model is obtained by collecting and training a sequence of images of the target scene”) (Maggio p. 1: “Neural Radiance Fields (NeRF) have gained significant popularity, as they can encode both 3D geometry and appearance of an environment”) (Lin p. 2: “NeRF [15] is a milestone approach demonstrating that neural
scene representations have the capabilities to synthesize photo-realistic views”). Lin et
al., “Parallel Inversion of Neural Radiance Fields for Robust Pose Estimation,” arXiv:2210.10108 (2022), is applied below; Maggio et al., “Loc-NeRF: Monte Carlo Localization using Neural Radiance Fields,” arXiv:2209.09050 (2022), is cited as a publication and listed on the PTO-892.
Outputting a camera pose as a positioning result for navigation (SPEC ¶ 3: “Currently,
satellites are needed in navigation to perform perspective positioning”) (Maggio p. 1:
“Vision-based localization is a foundational problem in robotics and computer vision,
with applications ranging from self-driving vehicles [1] to robot manipulation [2]”}.
Image retrieval from a scene-specific library and similarity-based candidate selection: Bao (U.S. 2022/0148302) acquires candidates from an image library by visual-word similarity scoring (¶125) and feature-matching-count ranking (¶123).
Coarse-to-fine search-space narrowing for pose estimation: Bao selects top candidates by global similarity and then refines within a local neighborhood of sequential frames around the best match (¶¶111-112).
Determining camera pose from matched image correspondences: Bao computes camera posture from feature correspondences using the Perspective-n-Point algorithm (¶¶113-114).
Generating candidate images at specified viewing angles using a trained model (claims 7, 16): Tang (CN 115409949 A) uses a pre-trained perspective-image generation model to output images at target viewpoints from source views (¶¶15-16, 45).
Viewing-angle-range-based image generation and iterative pose refinement (claims 4-6, 9, 13-15, 18): Xiao (CN 116309836 A) projects reference images from various angles through a sparse point cloud (¶¶93, 99) and iteratively optimizes the predicted pose until a preset condition is met (¶¶12, 113).
The ordered combination – coarse retrieval, neighborhood narrowing, fine retrieval, pose assignment – is the conventional image-based visual localization pipeline shown by these
references, and is the standard multi-stage practice those publications themselves describe (Maggio p. 1: classical approaches “typically address the task by adopting a multi-stage paradigm”; p. 3: “According to standard Monte Carlo localization”; p. 3: “As prescribed by
standard particle filtering, the resampling step allows retaining particles that are more likely to
correspond to good pose estimates while discarding less likely hypotheses”) (Lin p. 4: “We draw
inspiration from the particle filtering framework”) . The combination adds nothing significantly more than the abstract idea.
Claim 10.
Claim 10 recites the same operations by a generic processor and memory. Generic
computing hardware does not supply a practical application or an inventive concept. See Alice Corp. v. CLS Bank, 573 U.S. 208, 223 (2014). Rejected for the reasons given for claim 1.
Claims 2-9 and 11-18.
The dependents narrow or further specify the abstract idea without integrating it:
Claims 2, 11 – initial-image selection further specifies data gathering. Conventional per Bao (¶125).
Claims 3, 12 – similarity or object-matching variants further specify the comparison. Conventional per Bao (¶¶123, 125).
Claims 4-6, 13-15 – viewing-angle-range definition is a mathematical relationship. Angle-range sampling is conventional per Xiao (¶¶93, 99).
Claims 7, 16 – generic "first model" trained to output viewing-angle images is "apply it" with a conventional novel-view-synthesis model. Conventional per Tang (¶¶15-16), per Maggio p. 2 (a trained NeRF renders novel views from a pose), and per the specification's own use of a model pre-constructed for the scene (SPEC ¶¶ 24, 72).
Claims 8, 17 – quantitative relationship between candidate-set sizes.
Claims 9, 18 – iterative refinement on the abstract comparison. Conventional per Xiao (¶¶12, 113).
Claims 2-9 and 11-18 are rejected for the reasons given for claims 1 and 10.
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, 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.
The factual inquiries for establishing a background for determining obviousness under 35 U.S.C. 103 are summarized as follows:
1. Determining the scope and contents of the prior art.
2. Ascertaining the differences between the prior art and the claims at issue.
3. Resolving the level of ordinary skill in the pertinent art.
4. Considering objective evidence present in the application indicating obviousness or nonobviousness.
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.
Claims 1-3, 8, 10-12, and 17 is/are rejected under 35 U.S.C. 103 as being unpatentable over Bao et al., US 2022/0148302 A1 (“Bao”) in view of Lin et al., “Parallel Inversion of Neural Radiance Fields for Robust Pose Estimation,” (hereinafter “Lin”)..
Claims 1 and 10.
Bao discloses a data processing method (and corresponding device having a memory storing
computer programs and a processor coupled to the memory and configured to execute the
computer programs to perform the method) (¶¶251-252), comprising:
obtaining a first image collected from a target scene (BAO: "the first image being an image collected by a camera" (Claim 1; ¶259). This teaches the captured query image of the target scene);
obtaining a plurality of second images, each second image being an image corresponding to the target scene and not obtained by an image acquisition device for the target scene (BAO: “images corresponding to at least one same visual word with the first image in the image library are determined to obtain multiple primary images; and top Q % of images of which corresponding visual word vectors have a highest similarity with the visual word vector of the first image in the multiple primary images are determined to obtain the multiple candidate images” (¶125). “M images having the largest number of features matching with the first image in the multiple candidate images are acquired to obtain the first candidate image
sequence” (¶ 123); “the M is 5” (¶124); This teaches a plurality of second images corresponding to the target scene.);
determining a second target image from the plurality of second images according to the first image (BAO: "M images having the largest number of features matching with the first image in the multiple candidate images are acquired to obtain the first candidate image sequence" (¶123); "the M is 5" (¶124); the leading frame is "the first reference image frame" of the first candidate image sequence (¶¶100, 111). This teaches selecting the candidate most similar to the first image as the second target image.);
obtaining a plurality of third images, the plurality of third images being images corresponding to the target scene, and each third image meeting a first viewing angle condition with the second target image, wherein viewing angle data of each third image and viewing angle data of the second target image satisfy a viewing angle similarity condition (BAO: "a first posture of the camera is determined according to a first image sequence and the first image, the first image sequence including multiple successive image frames neighboring to a first reference image frame in the image library" (¶111); the first image sequence "includes front K1 image frames for the first reference image frame, the first reference image frame, and rear K1 image frames for the first reference image frame ... the K1 is 10" (¶112). This teaches a plurality of third images neighboring the second target image (the first reference image frame), the spatial/temporal-neighborhood relationship being a first viewing angle condition.);
determining a third target image from the plurality of third images according to the first image (BAO: "from features extracted from each image in the first image sequence, F features matching with features extracted from the first image are determined, the F being an integer greater than 0" (¶113). This teaches selecting, from the third images, the image-frame correspondences matching the first image as the third target image.) (Lin p. 4: “We keep the
first r ratio of the hypotheses with the lowest losses”; with the neighborhood rendered from
the radiance field as set out below, the rendering with the lowest loss against the first image is the third target image.); and
determining viewing angle data corresponding to the first image according to viewing angle data of the third target image (BAO: "the first posture is determined according to the F features, spatial coordinates corresponding to the F features in a point cloud map and internal parameters of the camera" (¶113); the first posture is determined “by using a Perspective-n-Point (PnP) algorithm” (¶114); " In a case where it is determined that the position of the camera is successfully localized according to the first posture, the first posture is determined as the target posture. " (¶111). This teaches outputting the camera pose (viewing angle data) of the first image based on the third target image's known pose in the point cloud map.) (Lin p. 1: “we can predict the translation and rotation of the camera”;
p. 1, Fig. 1: the “final (green)” hypothesis; the pose of the final lowest-loss rendering is the pose assigned to the observed image.) and
outputting the determined viewing angle data corresponding to the first image as a viewing
angle positioning result in a navigation application (BAO: “The method for visual localization
provided by the embodiments of the present disclosure can be applied to scenarios such as
position recognition, and position navigation” (¶ 84); “In 506, the server sends position
information to the terminal” (¶ 191); “The position information may include the 3D position of the camera, direction of the camera and sign information. The sign information indicates a
route through which the user walks from the current position to the destination address”
(¶ 192); “the server may accurately localize the position of the camera, and provides navigating information for the user” (¶ 195). This teaches outputting the determined camera posture as a positioning result in a navigation application.).
Bao teaches all of the subject matter as described above except for specifically teaching each second image being not obtained by an image acquisition device for the target scene; and
viewing angle data of each third image and viewing angle data of the second target image
satisfying a viewing angle similarity condition. However, Lin in the same field of endeavor (6-DoF camera pose estimation from a single RGB image against a model of the scene) teaches obtaining a plurality of candidate images of the target scene that are not obtained by an image acquisition device for the target scene (Lin p. 4: “we generate PN camera pose hypotheses around the start pose, with translation and rotation offsets uniformly sampled in the Euclidean space and SO(3), respectively”; Lin p. 2: the trained radiance field “can synthesize novel views by querying 5D coordinates along the camera rays and use classic volume rendering techniques to project the output colors and densities into an image”; the image rendered at each hypothesis pose is a candidate image of the scene produced by the network rather than captured by a camera), and obtaining the neighboring images at poses whose viewing angle data satisfy a viewing angle similarity condition with the retained candidate (Lin p. 4: “we generate PN camera pose hypotheses around the start pose, with translation and rotation offsets uniformly sampled in the Euclidean space and SO(3), respectively”; with the pose of Bao's retrieved frame as the start pose, each hypothesis is a camera pose, which is viewing angle data, sampled within a bounded range of that pose; Lin p. 4: “We keep the first r ratio of the hypotheses with the lowest losses and resample the remaining hypotheses around them with small offsets”).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to render Bao’s candidate images and the images neighboring the
retrieved frame from a radiance field trained for the scene as Lin does, with the pose of the
retrieved frame as Lin’s starting pose, in place of Bao’s stored camera frames, because Bao’s
neighborhood covers only viewpoints the camera visited (Bao ¶¶ 107, 198 library recorded, ¶ 112 adjacent-frame neighborhood), rendering it from Lin’s radiance field, which renders a view at any sampled pose (Lin p. 2), allows sampling at chosen poses around the retrieved frame, whose pose supplies the rough starting pose Lin's search requires (Lin p. 1). KSR lnt'I Co. v. Teleflex Inc., 550 U.S. 398 (2007). Lin reports its method on real scenes photographed with a phone (Lin p. 5: “complex real-world scenes captured with a handheld cellphone”), and Bao's query images are phone photographs (Bao ¶ 3: “surrounding rich visual information can be acquired by using a mobile phone and other devices for photographing images”). Applying Lin’s operation to Bao’s photographs requires no change in that operation.
Claims 2 and 11.
Bao and Lin teach the data processing method according to claim 1, wherein obtaining the plurality of second images comprises: obtaining a plurality of initial images, the plurality of initial images being images corresponding to the target scene and including at least one of the following: a plurality of randomly determined images or a plurality of images selected by preset information; and obtaining the plurality of second images according to the plurality of initial images and the first image (BAO: "images corresponding to at least one same visual word
with the first image in the image library are determined to obtain multiple primary images"
i.e., initial images selected by preset information (visual-word membership) "and top Q % of images of which corresponding visual word vectors have a highest similarity with the visual word vector of the first image in the multiple primary images are determined to obtain the multiple candidate images" (¶125). This teaches initial images selected by preset information, with the second images (the M images acquired from the candidate images by feature matching with the first image, ¶ 123) obtained from the initial images according to the first image.).
Claims 3 and 12.
Bao and Lin teach the data processing method according to claim 1, wherein determining the second target image from the plurality of second images according to the first image comprises: determining similarity information between the first image and each second image, and obtaining the second target image according to the similarity information (BAO: visual-word-vector similarity scoring (¶125) followed by feature-matching count quantification (¶123). This teaches similarity-based selection of the second target image.); or identifying object images in the first image and each second image, and obtaining the second target image according to matching situation between the object images in the first image and each second image.
Claims 8 and 17.
Bao and Lin teach the data processing method according to claim 1, wherein: the plurality of second images and the plurality of third images are obtained in a same manner (BAO: both drawn from the same image library and indexed by the same vocabulary-tree feature classification (¶¶125, 127)) (Lin p. 2: in the combination both sets are rendered by the same trained radiance field, which “can synthesize novel views by querying 5D coordinates along the camera rays and use classic volume rendering techniques to project the output colors and densities into an image”); and the number of the plurality of third images is greater than the number of the plurality of second images ((BAO: the first candidate image sequence contains "M is 5" images (¶124), while the first image sequence contains 2K1+1 successive image frames neighboring the first reference image frame, with K1= 10 yielding 21 frames (¶112). 21
>
5. This teaches the recited quantitative relationship.).
Claims 4, 5, 9, 13, 14, and 18 are rejected under 35 U.S.C. 103 as being unpatentable over Bao as applied to claims 1 and 10 above, and further in view of Xiao et al., CN 116309836 A ("Xiao").
Claims 4 and 13.
Bao teaches all of the subject matter as described above except for specifically teaching
Lin teaches this (Lin p. 4: “we generate PN camera pose hypotheses around the start pose, with translation and rotation offsets uniformly sampled in the Euclidean space and SO(3),
respectively”; the sampling range about the pose of the retrieved frame is a first viewing angle range, each sampled hypothesis is one of multiple first viewing angle data in that range, and the radiance field renders a third image at each; Lin p. 4: “resample the remaining hypotheses around them with small offsets”).
However, Xiao in the same field of endeavor (visual image-based three-dimensional object pose recognition) also teaches determining a first viewing angle range according to the viewing angle data of the second target image; determining multiple first viewing angle data in the first viewing angle range; and obtaining the plurality of third images according to the multiple first viewing angle data (XIAO: "The reference image can be projected from various angles through the sparse point cloud model" (¶93) and "the sparse 3D point cloud model of the target is projected from various angles to generate multiple reference images" (¶99). The target reference image (second target image) is identified, and a viewing-angle neighborhood around it is established by aligning the first image to the target reference image based on the relative plane rotation angle (¶101). The plurality of third images is then derived from this neighborhood as feature maps projected to a three-dimensional volume space around the target reference image's pose (¶¶104-105). This teaches the recited viewing-angle range and angle-based generation of third images.).
It would have been obvious before the effective filing date of the claimed invention to combine Bao with Xiao. Bao's sequential-frame neighborhood is geometrically coarse; Xiao's angle-based projection of the point cloud yields a denser sampling near the matched viewpoint and addresses Xiao's stated problem that coarse pose estimation alone "cannot provide enough three-dimensional information" (¶103), improving 6-DoF accuracy.
Claims 5 and 14.
Bao, Lin, and Xiao teach the data processing method according to claim 4, wherein: the plurality of second images are obtained according to a second viewing angle range; and determining the first viewing angle range according to the viewing angle data of the second target image includes: determining the first viewing angle range from the second viewing angle range according to the viewing angle data of the second target image, wherein the first viewing angle range is included in the second viewing angle range (XIAO: the reference images are generated by projecting the sparse 3D point cloud model from various angles spanning the broad coverage of the scene (¶¶93, 99), and the rough pose estimation network selects a target reference image whose viewing angle is closest to the first image (¶¶10, 94-98); the local feature-projection neighborhood around the target reference's viewpoint (first viewing angle range) is contained within the broader projection coverage (second viewing angle range) (¶104-105). This teaches inclusion of the first range within the second range.).
Claims 9 and 18.
Bao teaches iterative refinement through sequential candidate evaluation (BAO: "in a case where it is determined that the position of the camera is not successfully localized according to the first posture, a second posture of the camera is determined according to a second image sequence and the first image," the second reference image frame being "a next image frame or a previous image frame of the first reference image frame in the second candidate image sequence" (¶111); "the visual localization apparatus may sequentially use each image frame according to a chronological order. .. to localize the position of the camera" (¶120)) but does not specifically teach
However, Xiao in the same field of endeavor teaches conditional iteration based on rotational and positional relationships between the second target image and the first image (XIAO: the rough pose estimation network determines "the rotation information between the first image and the target reference image" and aligns the first image to the target reference
image "according to the rotation information" (¶¶10, 94-95); the fine pose estimation network then computes a "pose residual" between the predicted pose and an initial pose, and "the pose residual is combined with the initial pose to obtain a new pose prediction" (¶104-105); "the adjustment is repeated to realize iterative optimization of the predicted pose ... until a preset condition is reached" (¶12, ¶113), wherein the preset condition is the iteration count reaching a preset threshold (¶61, 144). This teaches the angle and position relationships and the iterative refinement recited in claims 9 and 18.).
It would have been obvious before the effective filing date of the claimed invention to combine Bao with Xiao. Xiao's rotation-then-residual decomposition addresses the rotational and translational components of pose error separately and answers its own stated shortcoming of single-stage coarse estimation (¶103), improving accuracy.
Claims 7 and 16 are rejected under 35 U.S.C. 103 as being unpatentable over Bao in view of Lin as applied to claims 1 and 10 above, and further in view of Tang et al., CN 115409949 A
("Tang").
Claims 7 and 16.
Bao teaches all of the subject matter as described above except for specifically teaching(Lin p. 4: "resample the remaining hypotheses around them with small offsets"; Lin p. 1: "an instant NGP model trained from multiple views of the target"; Lin p. 2: the trained radiance field "can synthesize novel views by querying SD coordinates along the camera rays and use classic volume rendering techniques to project the output colors and densities into an image"; the radiance field receives a camera pose and outputs the image of the scene from that pose).
However, Tang in the same field of endeavor (visual-angle image generation in a 3D scene) also teaches inputting viewing angle data into a model trained to output viewing angle images corresponding to the input viewing angle data (TANG: "obtaining multiple source
perspective images under multiple source perspectives in a preset 3D scene; according to
the plurality of source perspective images and the target perspective, a pre-trained
perspective image generation model is used to generate a target two-dimensional image
under the target perspective in the preset three-dimensional scene" (¶15-16); the model is trained such that "given several 2D source view images of a specific 3D scene and their corresponding camera poses, the new view generation task aims to generate a realistic 2D image corresponding to the scene under a new target view" (¶45); training is performed by minimizing reconstruction loss between the generated target image and a ground-truth image at the target perspective (¶¶69-70). This teaches a first model trained to output viewing angle images from input viewing angle data.).
It would have been obvious before the effective filing date of the claimed invention to combine Bao with Tang to populate Bao's candidate set with synthesized views at arbitrary viewing angles instead of relying on a pre-captured library. Tang explains the benefit: the model permits operation "when the number of given source view images is reduced" (¶4), reducing data collection burden and broadening angular coverage of the candidates.
Allowable Subject Matter
Claims 6 and 15 are objected to as being dependent upon a rejected base claim, but would be allowable if rewritten in independent form including all of the limitations of the base claim and any intervening claims; they remain rejected under 35 U.S.C. 101 as set forth above.
Conclusion
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 Ross Varndell whose telephone number is (571)270-1922. The examiner can normally be reached M-F, 9-5 EST.
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, O’Neal Mistry can be reached at (313)446-4912. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of an application may be obtained from the Patent Application Information Retrieval (PAIR) system. Status information for published applications may be obtained from either Private PAIR or Public PAIR. Status information for unpublished applications is available through Private PAIR only. For more information about the PAIR system, see https://ppair-my.uspto.gov/pair/PrivatePair. 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.
/Ross Varndell/Primary Examiner, Art Unit 2674
Prosecution Insights