Questions about individual parts of claim 1: what if that part was like this, and not that.
1.1Estimating content using optical flow
What if it was like thisWhat if the content estimation used dense optical flow instead of sparse motion vector predictors?
…and not this“estimating content for a current frame to be coded from motion vector predictors of previously coded reference frames,”
A dense optical flow algorithm computes motion vectors for every pixel block between the current frame and reference frames. The system uses block matching or gradient-based methods to generate a complete motion field without relying on existing coded motion vectors.
1.2Determining holes using depth maps
What if it was like thisWhat if the system determined content holes by analyzing depth maps instead of projection failures?
…and not this“determining, following the estimating, whether a content hole exists for the current frame, and”
The system projects depth maps from reference frames to the current frame's viewpoint. Pixels that do not receive depth values during this reprojection are marked as holes, identifying missing content regions before motion compensation.
1.3Deriving content using spatial interpolation
What if it was like thisWhat if the system derived hole content using spatial interpolation instead of intra-block derivation?
…and not this“when a content hole is determined to exist, on a processing block by processing block basis, deriving content for a processing block-sized portion of the content hole from other content of the respective processing block,”
The system applies bilinear or bicubic interpolation to pixels surrounding the hole boundary. It calculates missing pixel values by averaging or weighting the known pixel values from adjacent blocks outside the hole region.
1.4Processing blocks with variable sizes
What if it was like thisWhat if the processing block size varied dynamically based on hole geometry?
…and not this“wherein the processing block has a predetermined size.”
The system analyzes the shape and dimensions of each detected hole. It selects block sizes from a set of predefined dimensions, such as 8x8, 16x16, or 32x32, to match the hole boundaries and optimize processing efficiency.
1.1Estimating content using neural networks
What if it was like thisWhat if the content estimation used a trained neural network instead of motion vector predictors?
…and not this“estimating content for a current frame to be coded from motion vector predictors of previously coded reference frames,”
A convolutional neural network takes reference frames and camera parameters as input. The network outputs a predicted frame by learning spatial and temporal relationships, replacing traditional motion vector-based projection with learned feature extraction.
Section 2 · machine-generated
Variations
Further arrangements of the same subject matter, described here as of the publication date at the foot of this page.
Other field of use
motion field processing for medical imaging
The method estimates content for a current frame to be coded from motion vector predictors of previously coded reference frames. It determines whether a content hole exists for the current frame. When a content hole is determined to exist, on a processing block by processing block basis, it derives content for a processing block-sized portion of the content hole from other content of the respective processing block. The processing block has a predetermined size.
motion field processing for autonomous driving
The method estimates content for a current frame to be coded from motion vector predictors of previously coded reference frames. It determines whether a content hole exists for the current frame. When a content hole is determined to exist, on a processing block by processing block basis, it derives content for a processing block-sized portion of the content hole from other content of the respective processing block. The processing block has a predetermined size.
Component substitution
substituting optical flow for motion vectors
The method estimates content for a current frame to be coded from optical flow predictors of previously coded reference frames. It determines whether a content hole exists for the current frame. When a content hole is determined to exist, on a processing block by processing block basis, it derives content for a processing block-sized portion of the content hole from other content of the respective processing block. The processing block has a predetermined size.
Parameter or scale variation
processing block size of 16x16 pixels
The method estimates content for a current frame to be coded from motion vector predictors of previously coded reference frames. It determines whether a content hole exists for the current frame. When a content hole is determined to exist, on a processing block by processing block basis, it derives content for a processing block-sized portion of the content hole from other content of the respective processing block. The processing block has a predetermined size of 16x16 pixels.
Combination with existing systems
integrated with a video conferencing system
The method estimates content for a current frame to be coded from motion vector predictors of previously coded reference frames. It determines whether a content hole exists for the current frame. When a content hole is determined to exist, on a processing block by processing block basis, it derives content for a processing block-sized portion of the content hole from other content of the respective processing block. The processing block has a predetermined size. The method is integrated with a video conferencing system.
Automated or data-driven variant
automated hole filling using neural networks
The method estimates content for a current frame to be coded from motion vector predictors of previously coded reference frames. It determines whether a content hole exists for the current frame. When a content hole is determined to exist, on a processing block by processing block basis, it derives content for a processing block-sized portion of the content hole from other content of the respective processing block. The processing block has a predetermined size. The derivation of content is carried out by a trained neural network.
Section 3 · quoted from USPTO records
Other work
Passages reproduced word for word from two earlier-published U.S. patent documents, placed beside the language of claim 1. No characterization is offered. Each passage is marked A or B for the document it comes from; ¶ numbers are the paragraph numbers in the USPTO’s published full text of that document.
1.1estimating content for a current frame to be coded from motion vector predictors of previously coded reference frames,
The motion vectors of a target sample block are predicted from motion vectors of one or more previously coded sample blocks included in a co-located reference picture.
A · US 9,609,320 B2 · ¶ 47
When the prediction mode PredMode of the target coding block is the motion vector predictor mode in the inter prediction (MODE_INTER), the inter prediction unit 203 derives a plurality of motion vector predictor candidates using coding information of the previously decoded picture signal stored in the coding information storage memory 205 and registers the plurality of derived motion vector predictor candidates in the motion vector predictor candidate list to be described bel
B · US 11,503,277 B2 · ¶ 105
1.2determining, following the estimating, whether a content hole exists for the current frame, and
(B) judging whether the temporal motion vector prediction flag indicates (i) the temporal motion vector prediction is to be performed on the first picture or (ii) the temporal motion vector prediction is not to be performed on the first picture;
A · US 9,609,320 B2 · claim 1
1.3when a content hole is determined to exist, on a processing block by processing block basis, deriving content for a processing block-sized portion of the content hole from other content of the respective processing block,
1.4wherein the processing block has a predetermined size.
In the embodiment, a coding/decoding target picture is equally divided into units of a predetermined size.
B · US 11,503,277 B2 · ¶ 69
An empty cell means only that no passage was reproduced there.
Statements in these documents
Improvement of robustness is desired from such an image coding method and image decoding method.
A · US 9,609,320 B2 · ¶ 5
One non-limiting and exemplary embodiment provides an image coding method and an image decoding method that are capable of improving robustness.
A · US 9,609,320 B2 · ¶ 6
It should be noted that general and specific aspects described above may be implemented using a system, a method, an integrated circuit, a computer program, or a computer-readable recording medium such as a CD-ROM, or any combination of systems, methods, integrated circuits, computer programs, or computer-readable recording media.
A · US 9,609,320 B2 · ¶ 8
However, because the technology of Patent Document 1 involves a picture transform, there is a problem that the processing load is great.
B · US 11,503,277 B2 · ¶ 6
Cite aspatent.fyi, US 12,795,151 B2, claim 1, first published 2026-11-18T12:58:15-05:00, https://patent.fyi/US12795151/, SHA-256 13707ad948c7cb443b70a809d010b082c05d30087be90add13c1e8d18b4351be