Point Cloud Skip Decoding Using Reference Picture Prediction

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

The challenge of efficiently encoding and decoding point cloud data for transmission due to its large volume, particularly when using direct coding models, results in low coding performance.

Innovation Solution

Implementing a skip decode mode for point cloud decoding and a skip encode mode for point cloud encoding, where geometric information of current points is determined based on prediction points from reference pictures, reducing the need for direct encoding/decoding.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Device complexity

If direct coding model (DCM) is used to encode and decode current node geometric information, then complexity is greatly reduced, but coding performance becomes low

Engineering Contradiction:
Improveencoding complexityVSAvoidcoding performance
Core Design Contradiction:
Device complexityVSReliability

Solution Approach 1:

The patent applies preliminary action by determining in advance whether a current node satisfies the direct coding model conditions before encoding. The encoder pre-evaluates geometric relationships (such as whether the node is an isolated point or lies on a planar surface) and prepares appropriate encoding strategies beforehand, allowing efficient processing while maintaining coding performance through selective application of DCM versus alternative encoding methods

Inventive Principle:
Principle #10Preliminary action

2Device complexity

If direct coding model (DCM) is used for isolated points, then encoding complexity is reduced, but coding performance deteriorates

Engineering Contradiction:
Improveencoding complexityVSAvoidcoding performance
Core Design Contradiction:
Device complexityVSLoss of information

Solution Approach 1:

The patent applies local quality by differentiating encoding strategies based on the specific characteristics of each node. Instead of uniformly applying DCM to all isolated points, the encoder analyzes local geometric properties (such as spatial distribution patterns and contextual relationships with neighboring points) and selectively applies DCM only where appropriate, while using more sophisticated encoding methods for points where precision is critical

Inventive Principle:
Principle #3Local quality

Data Source

PatentUS20250392699A1Point cloud encoding method, point cloud decoding method, and storage medium
Publication Date: 2025.12.25 GUANGDONG OPPO MOBILE TELECOMMUNICATIONS CORP LTD
  • US20250392699A1 patent drawing
  • US20250392699A1 patent drawing
  • US20250392699A1 patent drawing

AI summary

A point cloud decoding method includes: determining a decoding mode of a current point in a current node, where the current node is a node to be decoded in a current picture to be decoded; in a case where the decoding mode of the current point is a skipDecodeMode, determining at least one prediction point of the current point among points included in N prediction nodes of the current node, where the prediction nodes are nodes corresponding to the current node in prediction reference pictures of the current picture to be decoded, the skipDecodeMode is a mode for skipping decoding of geometric information of the current point, and N is a positive integer; and determining the geometric information of the current point based on geometric information of the at least one prediction point.