# US 2026/0230645 A1 · Learning-based predictive coding for dynamic point clouds

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2026/0230645 A1
- Document ID: US20260230645A1
- Status: Published application
- Kind: A1
- Application number: 19/147,766
- Filed: 2024-01-03
- Published: 2026-08-06
- Earliest claimed: 2023-01-16
- Assignees: InterDigital VC Holdings, Inc.
- Applicants: InterDigital VC Holdings, Inc.
- Inventors: Junghyun AHN; Jiahao PANG; Dong TIAN
- Web page: https://patentgrep.com/patent/US20260230645A1

## CPC

- G06T 9/002: {using neural networks}
- H04N 19/597: specially adapted for multi-view video sequence encoding
- G06T 9/00: Image coding (bandwidth or redundancy reduction for static pictures H04N1/41; coding or decoding of static colour picture signals H04N1/64; methods or arrangements for coding, decoding, compressing or decompressing digital video signals H04N19/00)
- H04N 19/105: Selection of the reference unit for prediction within a chosen coding or prediction mode, e.g. adaptive choice of position and number of pixels used for prediction
- H04N 19/107: between spatial and temporal predictive coding, e.g. picture refresh
- H04N 19/137: Motion inside a coding unit, e.g. average field, frame or block difference
- H04N 19/159: Prediction type, e.g. intra-frame, inter-frame or bidirectional frame prediction
- H04N 19/176: the region being a block, e.g. a macroblock
- H04N 19/54: using feature points or meshes
- G06T: IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- H04N: PICTORIAL COMMUNICATION, e.g. TELEVISION

## Abstract

In one implementation, we introduce a dynamic point cloud compression (DPCC) architecture which learns how to code each spatial partition (block) based on the nature of the points in the block. A block of current frame that has high correlation to a block in the previous (reference) frame is defined as “inter mode block”. Otherwise, the block with less temporal correlation can be defined as an “intra mode block”. These blocks combined in one PC frame can improve the coding performance since it simplifies the packing of various residual features in a point cloud (PC) frame. Besides, it can also enhance the quality of reconstruction as a proper feature learning method can be assigned on each block. Moreover, the proposed two-level, block- and frame-level architecture for DPCC and the novel block-level motion compensation subtracted within the feature space help the decoder to more robustly synthesize points.

## Claims

### Claim 1

A method of decoding a point cloud frame in a dynamic point cloud sequence, comprising:
decoding motion information for a current point cloud block, wherein said current point cloud block is inter coded;
obtaining a reference point cloud block in a previously reconstructed point cloud frame for predicting said current point cloud block, based on said motion information for said current point cloud block;
obtaining a feature representative of said reference point cloud block, using at least a point-based neural network;
decoding residual feature information for said current point cloud block that represents a difference between said current point cloud block and said reference point cloud block;
obtaining a feature for said current point cloud block, based on said decoded residual feature information and said feature representative of said reference point cloud block; and
reconstructing positions of 3D points in said current point cloud block based on said feature for said current point cloud block, using at least another point-based neural network.

### Claim 2

(canceled)

### Claim 3 (depends on 1)

The method of claim 1, further comprising:
decoding information indicating positions of occupied blocks in said point cloud frame.

### Claim 4 (depends on 3)

The method of claim 3, further comprising:
decoding information indicating coding modes of respective occupied blocks in said point cloud frame.

### Claim 5 (depends on 1)

The method of claim 1, further comprising:
concatenating said residual feature information and said feature representative of said reference point cloud block; and
applying a series of convolutional layers to said concatenated feature to form said feature representative of said current point cloud block.

### Claim 6 (depends on 1)

The method of claim 1, wherein said at least another point-based neural network includes a series of MLP layers.

### Claim 7 (depends on 1)

The method of claim 1, wherein said decoding residual feature information for said current point cloud block comprises:
decoding a downsampled feature map, wherein said downsampled feature map is augmented by data indicating whether an intra or inter mode is used for a corresponding block in a current point cloud frame; and
upsampling and aggregating said downsampled feature map to obtain said residual feature information.

### Claim 8 (depends on 1)

The method of claim 1, wherein another block in a current point cloud frame is decoded in an intra mode without a reference point cloud frame.

### Claim 9

A method of encoding a point cloud frame in a dynamic point cloud sequence, comprising:
estimating motion information for a current point cloud block based on a current point cloud frame and a previously reconstructed point cloud frame;
obtaining a reference point cloud block from a previously reconstructed point cloud frame for predicting said current point cloud block, based on said motion information for said current point cloud block;
obtaining a first feature representative of said current point cloud block and a second feature representative of said reference point cloud block, using at least a point-based neural network;
obtaining a difference between said first feature representative of said current point cloud block and said second feature representative of said reference point cloud block to form a residual feature; and
encoding said residual feature.

### Claim 10

(canceled)

### Claim 11 (depends on 9)

The method of claim 9, further comprising:
encoding information indicating positions of occupied blocks in said point cloud frame.

### Claim 12 (depends on 11)

The method of claim 11, further comprising:
encoding information indicating coding modes of respective occupied blocks in said point cloud frame.

### Claim 13 (depends on 9)

The method of claim 9, wherein said obtaining a difference comprises:
concatenating said first feature representative of said current point cloud block and said second feature representative of said reference point cloud block; and
applying a series of convolutional layers to said concatenated feature to form said difference.

### Claim 14 (depends on 9)

The method of claim 9, wherein said first feature representative of said current point cloud block and said second feature representative of said reference point cloud block are obtained by a shared point-based neural network.

### Claim 15 (depends on 9)

The method of claim 9, further comprising:
downsampling and aggregating a feature map to form a downsampled feature map, wherein said feature map is augmented by data indicating whether an intra or inter mode is used for a corresponding block in said current point cloud frame; and
encoding said downsampled feature map.

### Claim 16 (depends on 9)

The method of claim 9, wherein another block in said current point cloud frame is encoded in an intra mode without a reference point cloud frame.

### Claim at position 17

17-18. (canceled)

### Claim 19

An apparatus for decoding a point cloud frame in a dynamic point cloud sequence, comprising one or more processors and at least one memory coupled to said one or more processors, wherein said one or more processors are configured to:
decode motion information for a current point cloud block, wherein said current point cloud block is inter coded;
obtain a reference point cloud block in a previously reconstructed point cloud frame for predicting said current point cloud block, based on said motion information for said current point cloud block;
obtain a feature representative of said reference point cloud block, using at least a point-based neural network;
decode residual feature information for said current point cloud block that represents a difference between said current point cloud block and said reference point cloud block;
obtain a feature for said current point cloud block, based on said decoded residual feature information and said feature representative of said reference point cloud block; and
reconstruct positions of 3D points in said current point cloud block based on said feature for said current point cloud block, using at least another point-based neural network.

### Claim 20 (depends on 19)

The apparatus of claim 19, wherein said one or more processors are further configured to:
concatenate said residual feature information and said feature representative of said reference point cloud block; and
apply a series of convolutional layers to said concatenated feature to form said feature representative of said current point cloud block.

### Claim 21 (depends on 19)

The apparatus of claim 19, wherein said at least another point-based neural network includes a series of MLP layers.

### Claim 22

An apparatus for encoding a point cloud frame in a dynamic point cloud sequence, comprising one or more processors and at least one memory coupled to said one or more processors, wherein said one or more processors are configured to:
estimate motion information for a current point cloud block based on a current point cloud frame and a previously reconstructed point cloud frame;
obtain a reference point cloud block from a previously reconstructed point cloud frame for predicting said current point cloud block, based on said motion information for said current point cloud block;
obtain a first feature representative of said current point cloud block and a second feature representative of said reference point cloud block, using at least a point-based neural network;
obtain a difference between said first feature representative of said current point cloud block and said second feature representative of said reference point cloud block to form a residual feature; and
encode said residual feature.

### Claim 23 (depends on 22)

The apparatus of claim 22, wherein said one or more processors are further configured to:
concatenate said first feature representative of said current point cloud block and said second feature representative of said reference point cloud block; and
apply a series of convolutional layers to said concatenated feature to form said difference.

### Claim 24 (depends on 22)

The apparatus of claim 22, wherein said first feature representative of said current point cloud block and said second feature representative of said reference point cloud block are obtained by a shared point-based neural network.
