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 .
Information Disclosure Statement
The information disclosure statement(s) (IDS) submitted on 10/11/2024, 01/27/2026 & 03/12/2026 is/are in compliance with the provisions of 37 CFR 1.97. Accordingly, the information disclosure statement(s) is/area being considered by the examiner.
Disposition of the Claims
In response to applicant’s Election of Invention II, received on 08/04/2026, all requested changes to the claims have been entered.
Claim(s) 1-30 were previously pending.
No Claim(s) have been added.
Claim(s) 1-17 have been withdrawn.
Claim(s) 18-30 are currently pending.
Response to Arguments
Election with Traverse
Applicant elects Invention II (claims 18-30) with traverse, on the grounds that there would not be a serious search burden between the claims of Inventions I & II.
The examiner respectfully disagrees with this conclusion. As evidenced in the subsequent rejections under 35 U.S.C. § 103, Invention II is directed primarily towards disparity estimation. While some of the dependent claims of Invention II recite utilizing selectivity kernels that would overlap with the subject matter of Invention I (namely, claims 21-25), the realm of search is sufficiently distinct, as Invention I is primarily dedicated to architecture of the selectivity kernels, with little emphasis on their use for disparity estimation (outside of claim 15). Search for Inventions I & II would likely necessitate distinct prior art references to teach the independent claim of each invention and its associated dependent claims.
Furthermore, in the IDS, filed 01/27/2026 & 03/12/2026, a partial and complete International Search Report for PCT/US2025/041481 is provided. Here, the international searching authority explicitly considers the original claims (which recite substantially identical subject matter to the claims as currently presented) to embody two distinct inventions. The lack of unity analysis between the two claimed inventions under Rule 13.1 & 13.2 of the Patent Cooperation Treaty require more stringent analysis to demonstrate lack of unity compared to U.S. restriction practice outlined in MPEP § 800. This further supports the examiner’s original requirement for restriction of Inventions I & II.
For at least these reasons, the Restriction Requirement, filed 06/05/2026 will be maintained.
Claim Rejections - 35 USC § 103
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis (i.e., changing from AIA to pre-AIA ) for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
Claim(s) 18, 20, 26, 28 & 29 is/are rejected under 35 U.S.C. 103 as being unpatentable over Xie et al ("Feature-Guided Spatial Attention Upsampling for Real-Time Stereo Matching Network", 2020, IEEE), hereinafter referred to as “Xie”, in view of Zhao et al (US 2014/0153639 A1), hereinafter referred to as “Zhao”.
Regarding claim 18, Xie teach An apparatus (a 2080 Ti GPU [pg. 43; Sec: Implementation Details - 02]) configured to perform disparity estimation (Xie disclose an end-to-end real-time stereo matching (i.e., disparity estimation) network, deemed RTSMNet [pg. 38; Sec: Abstract]), comprising:
and
one or more processors, coupled to the one or more memories (2080 Ti GPU [pg. 43; Sec: Implementation Details - 02] – the examiner notes a GPU inherently comprises one or more processors and one or more memories), configured to:
determine to perform (given the use of the disjunctive “or” only one of the listed limitations requires mapping) or disparity upsampling based disparity estimation of the first scene based on the first complexity of the first scene (RTSMNet uses feature-guided spatial attention upsampling (FSAU) modules to create a full resolution left-right disparity map to recover thin structures and detailed information from a scene depicted in a left and right image [pg. 39, col. 1 - ¶03; Fig. 1], the FSAU module utilizes global and local features (i.e., a first complexity / information indicating the complexity of a scene) from the global and local feature extraction (GLFE) module provided in a guided feature map to guide the upscaling of low resolution features [pg. 40, Sec: Global and Local Feature Extraction - ¶01-03 & Sec: FSAU Module - ¶01-02]); and
perform disparity estimation of the first scene (RTSMNet uses feature-guided spatial attention upsampling (FSAU) modules to create a full resolution left-right disparity map to recover thin structures and detailed information a scene depicted in a left and right image [pg. 39, col. 1 - ¶03; Fig. 1])
Xie, while accounting for scene complexity using upsampling based disparity estimation, fails to teach an explicit step of determining scene complexity. Zhao, on the other hand, is analogous art pertinent to the field of endeavor of the present application and disclose a video encoding device that captures the complexity of a scene to determine how to encode high and low complexity scenes. Zhao teach one or more memories (Zhao: memory module 232 of the video encoding/decoding system 102 [¶0016; Fig. 2]) configured to store first information that indicates first complexity of a first scene; (Zhao: a group of picture (GOP) adaptation module 150 generates GOP adaptation data (i.e., first information) indicating scene complexity [¶0013; Fig. 2], which is further detailed in steps 500-504 of Fig. 5 [¶0029-30]).
Zhao explains that their device uses scene complexity to determine when to utilize hierarchical B frames: during high complexity scenes, there is a small benefit but a great degree of computational cost, but in low complexity scenes, the benefit is greater with an overall lower computational cost. This enables the device to utilize scene complexity to determine when to perform computationally intensive processes to optimize overall memory bandwidth usage [¶0013]. One of ordinary skill in the art would be motivated to utilize the complexity determination taught by Zhao, but instead of using it to inform the use of hierarchical B frames for video encoding, to instead use that determination inform the feature-guided spatial upsampling module of Xie. This would facilitate Xie’s disparity estimation procedure to dynamically decide to perform upsampling based disparity estimation only in scenes that have low complexity, minimizing upsampling performed on high complexity scenes to arrive at the invention of the instant application.
Turning to claim 20, Xie in view of Zhao teach The apparatus of claim 18 (described above), wherein the first complexity of the first scene is based on one or more of:
a level of motion in the first scene (Zhao: the GOP adaptation module 150 detects high or low scene complexity in a sequence of images, wherein scenes with uniform, simple motion for objects is considered low complexity [¶0019]); or
(given the use of the disjunctive “or” only one of the listed limitations requires mapping).
Zhao again explains that their device uses scene complexity to determine when to utilize hierarchical B frames: during high complexity scenes, there is a small benefit but a great degree of computational cost, but in low complexity scenes, the benefit is greater with an overall lower computational cost. This enables the device to utilize scene complexity to determine when to perform computationally intensive processes to optimize overall memory bandwidth usage [¶0013]. One of ordinary skill in the art would be motivated to utilize the complexity determination taught by Zhao, but instead of using it to inform the use of hierarchical B frames for video encoding, to instead use that determination inform the feature-guided spatial upsampling module of Xie. This would facilitate Xie’s disparity estimation procedure to dynamically decide to perform upsampling based disparity estimation only in scenes that have low complexity (such as scenes with low motion), minimizing upsampling performed on high complexity scenes to arrive at the invention of the instant application.
As for claim 26, Xie in view of Zhao teach The apparatus of claim 18 (explained previously), wherein the one or more processors are configured to:
determine to perform the (given the use of the disjunctive “or” only one of the listed limitations requires mapping) or the disparity upsampling based disparity estimation of the second scene based on the second complexity of the second scene (Xie: RTSMNet uses feature-guided spatial attention upsampling (FSAU) modules to create a full resolution left-right disparity map to recover thin structures and detailed information from a scene depicted in a left and right image [pg. 39, col. 1 - ¶03; Fig. 1], the FSAU module utilizes global and local features (i.e., a second complexity / information indicating the complexity of a scene) from the global and local feature extraction (GLFE) module provided in a guided feature map to guide the upscaling of low resolution features [pg. 40, Sec: Global and Local Feature Extraction - ¶01-03 & Sec: FSAU Module - ¶01-02] – the examiner notes that RTSMNet is capable of performing disparity estimation across a plurality of scenes); and
perform the disparity estimation of the second scene (Xie: RTSMNet uses feature-guided spatial attention upsampling (FSAU) modules to create a full resolution left-right disparity map to recover thin structures and detailed information a scene depicted in a left and right image [pg. 39, col. 1 - ¶03; Fig. 1])
Xie, again, fails to disclose an explicit step of receiving and storing complexity information of a scene, instead intrinsically guiding upsampling based on global and local features that indicate scene complexity. Zhao, however, teach receive second information that indicates a second complexity of a second scene; (Zhao: a group of picture (GOP) adaptation module 150 generates GOP adaptation data (i.e., first information) indicating scene complexity [¶0013; Fig. 2], which is further detailed in steps 500-504 of Fig. 5 [¶0029-30] – the examiner notes that Zhao’s GOP adaptation module is able to capture data indicating scene complexity across a plurality of scenes).
Zhao again explains that their device uses scene complexity to determine when to utilize hierarchical B frames: during high complexity scenes, there is a small benefit but a great degree of computational cost, but in low complexity scenes, the benefit is greater with an overall lower computational cost. This enables the device to utilize scene complexity to determine when to perform computationally intensive processes to optimize overall memory bandwidth usage [¶0013]. One of ordinary skill in the art would be motivated to utilize the complexity determination taught by Zhao, but instead of using it to inform the use of hierarchical B frames for video encoding, to instead use that determination inform the feature-guided spatial upsampling module of Xie. This would facilitate Xie’s disparity estimation procedure to dynamically decide to perform upsampling based disparity estimation only in scenes that have low complexity, minimizing upsampling performed on high complexity scenes to arrive at the invention of the instant application.
With respect to claim 28, Xie in view of Zhao teach The apparatus of claim 18 (as described previously), wherein to determine to perform (given the use of the disjunctive “or” only one of the listed limitations requires mapping) or the disparity upsampling based disparity estimation (RTSMNet uses feature-guided spatial attention upsampling (FSAU) modules to create a full resolution left-right disparity map to recover thin structures and detailed information a scene depicted in a left and right image [pg. 39, col. 1 - ¶03; Fig. 1]) is further based on at least one of:
(given the use of the disjunctive “or” only one of the listed limitations requires mapping); or
a computational efficiency (Xie: RTSMNet avoids too much computational burden by utilizing a low-resolution cost volume and employs five 3D convolutions for filtering [pg. 39, col. 1 - ¶03]).
Concerning claim 29, Xie in view of Zhao teach The apparatus of claim 18 (as previously explained), further comprising a modem, coupled to one or more antennas, and coupled to the one or more processors (Zhao: the video processing system 125 includes a receiving module 100, which can be a satellite broadcast receiver, a broadcast modem or 3G transceiver or the like [¶0010; Fig. 1] – the examiner notes that one of ordinary skill would understand that at least a 3G transceiver would at least have an RF antenna for communicating radio signals, and a modem for translating analog to digital signals and vice versa), wherein the modem and the one or more antennas are configured to receive the first information that indicates the first complexity of the first scene (Zhao: the GOP adaptation module 150 generates GOP adaptation data indicating scene complexity as part of the video encoding/decoding device 102 [¶0013; Fig. 2] – while the disclosure describes first obtaining video data via receiving module 100 (which may be a 3G transceiver comprising a modem and an antenna), and then utilizing that video data to generate information indicating a complexity of a scene, one of ordinary skill in the art would recognize that the receiving module could be reconfigured to also receive complexity information).
Zhao again explains that their device uses scene complexity to determine when to utilize hierarchical B frames: during high complexity scenes, there is a small benefit but a great degree of computational cost, but in low complexity scenes, the benefit is greater with an overall lower computational cost. This enables the device to utilize scene complexity to determine when to perform computationally intensive processes to optimize overall memory bandwidth usage [¶0013]. One of ordinary skill in the art would be motivated to utilize the complexity determination taught by Zhao, but instead of using it to inform the use of hierarchical B frames for video encoding, to instead use that determination inform the feature-guided spatial upsampling module of Xie. This would facilitate Xie’s disparity estimation procedure to dynamically decide to perform upsampling based disparity estimation only in scenes that have low complexity, minimizing upsampling performed on high complexity scenes to arrive at the invention of the instant application.
Claim(s) 19 & 27 is/are rejected under 35 U.S.C. 103 as being unpatentable over Xie et al ("Feature-Guided Spatial Attention Upsampling for Real-Time Stereo Matching Network", 2020, IEEE), hereinafter referred to as “Xie”, in view of Zhao et al (US 2014/0153639 A1), hereinafter referred to as “Zhao”, further in view of Kendall et al ("End-to-End Learning of Geometry and Context for Deep Stereo Regression", 2017, ArXiv), hereinafter referred to as “Khamis”.
Regarding claim 19, Xie in view of Zhao teach The apparatus of claim 18 (as previously taught), wherein to determine to perform the standalone disparity estimation or the disparity upsampling based disparity estimation, the one or more processors are configured to:
and
determine to perform the disparity upsampling based disparity estimation (Xie: RTSMNet uses feature-guided spatial attention upsampling (FSAU) modules to create a full resolution left-right disparity map to recover thin structures and detailed information from a scene depicted in a left and right image [pg. 39, col. 1 - ¶03; Fig. 1]).
Xie, however, is silent on comparing scene complexity to determine a type of upsampling to perform. Zhao, however, teach compare the first complexity of the first scene to a threshold (Zhao: the GOP adaptation module 150 compares scene complexity to a scene complexity threshold [¶0020]);
if the first complexity satisfies the threshold (Zhao: when the scene complexity is favorable (i.e., less than a scene complexity threshold), the scene is deemed to have low complexity [¶0020]),
if the first complexity does not satisfy the threshold, (Zhao: when the scene complexity is unfavorable (i.e., greater than a scene complexity threshold), the scene is deemed to have high complexity [¶0020]).
Zhao again explains that their device uses scene complexity to determine when to utilize hierarchical B frames: during high complexity scenes, there is a small benefit but a great degree of computational cost, but in low complexity scenes, the benefit is greater with an overall lower computational cost. This enables the device to utilize scene complexity to determine when to perform computationally intensive processes to optimize overall memory bandwidth usage [¶0013]. One of ordinary skill in the art would be motivated to utilize the complexity determination taught by Zhao, but instead of using it to inform the use of hierarchical B frames for video encoding, to instead use that determination inform the feature-guided spatial upsampling module of Xie. This would facilitate Xie’s disparity estimation procedure to dynamically decide to perform upsampling based disparity estimation only in scenes that have low complexity, minimizing upsampling performed on high complexity scenes to arrive at the invention of the instant application.
Xie in view of Zhao, however, fail to explicitly disclose performing standalone disparity estimation without upsampling. Kendall, per contra, is analogous art pertinent to the field of endeavor and disclose disparity estimation deep learning architecture without upsampling. Kendall teach determine to perform the standalone disparity estimation; (Kendall: a end-to-end deep stereo regression architecture, GC-Net (Geometry and Context Network) directly regresses disparity from stereo-image pairs of a particularly scene producing a dense disparity estimate at the original input dimensions [Sec 1. Introduction - ¶04-05; Fig. 1 & Sec: 3. Learning End-to-end Disparity Regression - ¶01-02]).
Kendall disclose the advantages of their disparity estimation architecture, explaining that by incorporating wider contexts of the scene, their method handles challenging scenarios involving reflective surfaces, thin or texture-less surfaces, and large occlusions [Sec 4.2. KITTI Benchmark - ¶01-03; Fig. 4]; these challenging scenarios all highlight elements contributing to scene complexity. One of ordinary skill in the art would recognize utilizing the context-aware, standalone disparity estimation of Kendall with the base architecture of Xie in view of Zhao that utilizes scene complexity to guide upsampling-based disparity estimation would enable the overall architecture to selectively employ disparity estimation methods based on a degree of scene complexity to arrive at the invention of the instant application.
Concerning claim 27, Xie in view of Zhao teach The apparatus of claim 26 (as described previously), wherein:
and
to perform the disparity estimation of the second scene, the one or more processors are configured to perform the disparity upsampling based disparity estimation for the second scene (Xie: RTSMNet uses feature-guided spatial attention upsampling (FSAU) modules to create a full resolution left-right disparity map to recover thin structures and detailed information from a scene depicted in a left and right image [pg. 39, col. 1 - ¶03; Fig. 1]).
As stated before, Xie in view of Zhao fail to teach explicitly performing standalone disparity estimation without an upsampling operation. In contrast, Kendall teach to perform the disparity estimation of the first scene, the one or more processors are configured to perform the standalone disparity estimation for the first scene, (Kendall: GC-Net directly regresses disparity from stereo-image pairs of a particularly scene producing a dense disparity estimate at the original input dimensions [Sec 1. Introduction - ¶04-05; Fig. 1 & Sec: 3. Learning End-to-end Disparity Regression - ¶01-02]).
Kendall, again, disclose the advantages of their disparity estimation architecture, explaining that by incorporating wider contexts of the scene, their method handles challenging scenarios involving reflective surfaces, thin or texture-less surfaces, and large occlusions [Sec 4.2. KITTI Benchmark - ¶01-03; Fig. 4]; these challenging scenarios all highlight elements contributing to scene complexity. One of ordinary skill in the art would recognize utilizing the context-aware, standalone disparity estimation of Kendall with the base architecture of Xie in view of Zhao that utilizes scene complexity to guide upsampling-based disparity estimation would enable the overall architecture to selectively employ disparity estimation methods based on a degree of scene complexity to arrive at the invention of the instant application.
Claim(s) 21-25 is/are rejected under 35 U.S.C. 103 as being unpatentable over Xie et al ("Feature-Guided Spatial Attention Upsampling for Real-Time Stereo Matching Network", 2020, IEEE), hereinafter referred to as “Xie”, in view of Zhao et al (US 2014/0153639 A1), hereinafter referred to as “Zhao”, further in view of Kauff et al (US 2014/0270485 A1), hereinafter referred as “Kauff”.
As for claim 21, Xie in view of Zhao teach The apparatus of claim 18 (previously described), wherein:
to perform the disparity estimation, the one or more processors are configured to perform the disparity upsampling based disparity estimation (Xie: RTSMNet uses feature-guided spatial attention upsampling (FSAU) modules to create a full resolution left-right disparity map to recover thin structures and detailed information from a scene depicted in a left and right image [pg. 39, col. 1 - ¶03; Fig. 1]) using a machine learning model comprising a plurality of selectivity kernels, (Xie: RTSMNet is a neural network architecture that generates and utilizes adaptive kernels to obtain spatial-context of different areas to inform attention weighting [pg. 39, col. 2 - ¶01])
While Xie disclose adaptive kernels used to distinguish spatial clues from different areas, they fail to that each of the plurality of kernels performs a distinct type of selectivity. Kauff, on the other hand, is analogous art pertinent to the field of endeavor and disclose a plurality of filtering kernels to assist spatio-temporal disparity map estimation. Kauff teach and
each selectivity kernel, of the plurality of selectivity kernels, is configured to perform a different type of selectivity (Kauff: a distance kernel [¶0073-76; Fig. 5], confidence kernel [¶0079-86; Fig. 6], and temporal kernel [¶0087-88] are used to execute particular types of selectivity for stereo matching consistency).
Kauff further explain that their invention overcomes drawbacks of conventional bilateral filter kernels by accounting for confidence measures and spatio-temporal consistency across a scene during disparity estimation [¶0024-32]. One of ordinary skill in the art would recognize that by implementing the plurality of Kauff’s kernels to apply consistency across different scene modalities with the overall upsampling-based disparity estimation architecture of Xie in view of Zhao, one would increase the overall accuracy and robustness in the final disparity map, accounting for complimentary sources of scene reliability.
Turning to claim 22, Xie in view of Zhao, further in view of Kauff teach The apparatus of claim 21 (as described above), wherein:
a first selectivity kernel of the plurality of selectivity kernels is configured to perform geometric consistency selectivity based on a geometric consistency kernel (GCK) (Kauff: a distance kernel (interpreted as a geometric consistency kernel) adapts weighting via a distance function that calculating all possible paths between two pixel points, minimizing the influence of pixels in the filter window not belonging to the same object, avoiding misalignment of left and right images [¶0073-78; Fig. 5; eqs. 2 & 3] – the examiner notes the distance kernel performs geometric consistency selectivity by controlling for pixels belonging to the same object, ensuring consistent geometric relationships across a scene), and
a second selectivity kernel of the plurality of selectivity kernels is configured to perform temporal consistency selectivity based on a spatiotemporal selectivity kernel (Kauff: a temporal kernel is employed to enforce temporal consistency to control the influence of temporally adjacent frames, which is applied to the motion compensated disparity maps, restricted temporal filtering to reliably motion compensated (RMC) pixels, enforcing both temporal and spatial selectivity [¶0087]).
Kauff again explain that their invention overcomes drawbacks of conventional bilateral filter kernels by accounting for confidence measures and spatio-temporal consistency across a scene during disparity estimation [¶0024-32]. One of ordinary skill in the art would recognize that by implementing the plurality of Kauff’s kernels to apply consistency across different scene modalities with the overall upsampling-based disparity estimation architecture of Xie in view of Zhao, one would increase the overall accuracy and robustness in the final disparity map, accounting for complimentary sources of scene reliability.
Considering claim 23, Xie in view of Zhao, further in view of Kauff teach The apparatus of claim 22 (explained above), wherein a third selectivity kernel of the plurality of selectivity kernels is configured to perform estimation confidence selectivity based on an estimation confidence kernel (ECK) (Kauff: a confidence kernel is utilized to account for the confidence and reliability measures for disparity matching, enabling greater weighting to be used for reliable areas with high confidence and low weighting for areas with low reliability [¶0079-82; Fig. 6; eqs. 4-6]).
Kauff again explain that their invention overcomes drawbacks of conventional bilateral filter kernels by accounting for confidence measures and spatio-temporal consistency across a scene during disparity estimation [¶0024-32]. One of ordinary skill in the art would recognize that by implementing the plurality of Kauff’s kernels to apply consistency across different scene modalities with the overall upsampling-based disparity estimation architecture of Xie in view of Zhao, one would increase the overall accuracy and robustness in the final disparity map, accounting for complimentary sources of scene reliability.
With respect to claim 24, Xie in view of Zhao, further in view of Kauff teach The apparatus of claim 21 (as previously described), wherein:
to perform the disparity estimation of the first scene based on the determination for the first scene, the one or more processors are configured to obtain output data, that corresponds to input data of the first scene, from a machine learning model (Xie: left and right images (i.e. input data), are fed through RTSMNet’s GLFE, IDE, and FSAU modules to provide a final high resolution upsampled disparity map (i.e., output data of the scene depicted in the left and right images [Fig. 1]),
the output data is based on a composite of outputs from the plurality of selectivity kernels (Xie: the final high resolution upsampled disparity map is generated using a plurality of adaptive kernels to provide greater weighting from reliable areas [pg. 39, col. 2 - ¶01; Fig. 1]).
Turning to claim 25, Xie in view of Zhao, further in view of Kauff teach The apparatus of claim 24 (as described above), wherein the plurality of selectivity kernels are configured to perform selectivity based on at least one of:
(given the use of the disjunctive “or” only one of the listed limitations requires mapping); or
estimation confidence (Kauff: a confidence kernel is utilized to account for the confidence and reliability measures for disparity matching, enabling greater weighting to be used for reliable areas with high confidence and low weighting for areas with low reliability [¶0079-82; Fig. 6; eqs. 4-6]).
Kauff again explain that their invention overcomes drawbacks of conventional bilateral filter kernels by accounting for confidence measures and spatio-temporal consistency across a scene during disparity estimation [¶0024-32]. One of ordinary skill in the art would recognize that by implementing the plurality of Kauff’s kernels to apply consistency across different scene modalities with the overall upsampling-based disparity estimation architecture of Xie in view of Zhao, one would increase the overall accuracy and robustness in the final disparity map, accounting for complimentary sources of scene reliability.
Claim(s) 30 is/are rejected under 35 U.S.C. 103 as being unpatentable over Xie et al ("Feature-Guided Spatial Attention Upsampling for Real-Time Stereo Matching Network", 2020, IEEE), hereinafter referred to as “Xie”, in view of Zhao et al (US 2014/0153639 A1), hereinafter referred to as “Zhao”, further in view of Huang et al (US 2014/0270485 A1), hereinafter referred as “Huang”.
Lastly, for claim 30, Xie in view of Zhao teach The apparatus of claim 29 (as previously explained), however, Xie in view of Zhao fail to teach integrating a modem and one or more antennas into a vehicle, extra-reality device, or a mobile device.
Huang, on the other hand, is analogous art pertinent to the field of endeavor of the present application and describe a remote control system for partial control of a vehicle through a virtual reality headset. Huang teach wherein the modem and the one more antennas are integrated into at least one of:
a vehicle (Huang: a vehicle 102 includes one or more modems and one more antennas [¶0045]);
(given the use of the disjunctive “or” only one of the listed limitations requires mapping).
Huang further explain that their hardware accelerators 714 may include programmable vision accelerator(s) (PVA), to accelerate computer vision algorithms, citing the importance of rapid computation to be used for real-time stereo matching needed for autonomous vehicle operation to inform pedestrian recognition and lane detection [¶0197, 205-206]. One of ordinary skill in the art would recognize the benefit of integrating the overall complexity-based stereo estimation method taught by Xie in view of Zhao into the autonomous vehicle operation of Huang for real-time stereo matching to aid important computer vision tasks like pedestrian recognition and lane detection.
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure:
Nam et al (US 2025/0365389 A1) disclose an image encoding/decoding method that calculates image complexity based on pixel gradients to determine a final image resolution
Price et al (US 2021/0358155 A1) describe a method for calculating disparity by applying temporal consistency
Smirnov et al (US 2014/0063188 A1) outline a process for disparity estimation by downsampling image pairs first before generating a disparity map.
Khamis et al (“StereoNet: Guided Hierarchical Refinement for Real-Time Edge-Aware Depth Prediction”, 2018, arXiv) present a real-time stereo matching architecture that employs edge-aware upsampling to achieve high-quality outputs.
Ye et al (“Stereo Matching Method with Integrated Geometric Encoding for Disparity Refinement”, Jun 2024, IEEE International Joint Conference on Neural Network) provide a neural network-based stereo matching algorithm utilizing global and non-local geometric feature information to inform rapid disparity matching.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to Michael M. Sofroniou whose telephone number is (571)272-0287. The examiner can normally be reached M-F: 8:30 AM - 5:00 PM.
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, John M. Villecco can be reached at (571) 272-7319. 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.
/MICHAEL M SOFRONIOU/Examiner, Art Unit 2661 /JOHN VILLECCO/Supervisory Patent Examiner, Art Unit 2661
