K-D Tree Point Cloud Encoding via Deviation Quantization

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing methods for encoding and decoding point clouds are inefficient in terms of storage, bandwidth, and processing power, as they do not effectively minimize code complexity and compression time while maintaining data accuracy.

Innovation Solution

The use of a k-D tree algorithm for recursive partitioning of geometric constructs, combined with quantization and entropy encoding, to reduce the representation of point clouds into b-bit integer values and efficiently store and communicate geometric data.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If traditional encoding methods are used for point clouds, then data accuracy is maintained, but storage space and bandwidth requirements increase

Engineering Contradiction:
Improvedata accuracyVSAvoidstorage space
Core Design Contradiction:
Measurement precisionVSQuantity of substance

Solution Approach 1:

The point cloud data is segmented into multiple partitions using a k-D tree structure, dividing the geometric construct into hierarchical regions. This segmentation allows selective encoding of partition boundaries and point deviations rather than storing all raw coordinates, reducing storage requirements while maintaining reconstruction accuracy

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The encoding method transforms the representation parameters from storing complete coordinate values to storing only deviation values from partition boundaries. This parameter change reduces the quantity of data needed to represent the point cloud while preserving the ability to reconstruct the original geometry with acceptable accuracy

Inventive Principle:
Principle #35Parameter changes

2Loss of information

If traditional encoding methods are used for point clouds, then data completeness is preserved, but bandwidth requirements increase

Engineering Contradiction:
Improvedata completenessVSAvoidbandwidth
Core Design Contradiction:
Loss of informationVSQuantity of substance

Solution Approach 1:

The method extracts only the essential information needed for reconstruction by storing partition boundary definitions and point deviations rather than complete coordinate data. This extraction reduces the amount of data that must be transmitted over bandwidth-constrained channels while preserving the ability to reconstruct the point cloud

Inventive Principle:
Principle #2Taking out (Extraction)

Solution Approach 2:

The representation parameters are changed from full coordinate storage to deviation-based encoding, where only the differences from known partition boundaries are stored and transmitted. This parameter transformation reduces bandwidth requirements while maintaining data completeness for reconstruction purposes

Inventive Principle:
Principle #35Parameter changes

3Productivity

If complex encoding algorithms are used, then compression efficiency improves, but code complexity increases

Engineering Contradiction:
Improvecompression efficiencyVSAvoidcode complexity
Core Design Contradiction:
ProductivityVSDevice complexity

Solution Approach 1:

The complex encoding problem is segmented into manageable steps: k-D tree construction, partition boundary identification, deviation calculation, and entropy encoding. This segmentation of the encoding process into discrete, well-defined stages improves compression efficiency while keeping each individual step relatively simple and well-structured

Inventive Principle:
Principle #1Segmentation

4Manufacturing precision

If detailed point cloud representation is used, then geometric precision is maintained, but processing power requirements increase

Engineering Contradiction:
Improvegeometric precisionVSAvoidprocessing power
Core Design Contradiction:
Manufacturing precisionVSPower

Solution Approach 1:

The k-D tree structure and partition boundaries are pre-computed and stored before the actual encoding of point deviations. This preliminary action organizes the geometric data in advance, allowing the encoding step to focus only on computing simple deviation values from known boundaries, thereby maintaining geometric precision while reducing processing power requirements during encoding

Inventive Principle:
Principle #10Preliminary action

Data Source

PatentEP3324310B1K-d tree encoding for point clouds using deviations
Publication Date: 2020.09.09 GOOGLE LLC
  • EP3324310B1 patent drawingFigure 1
  • EP3324310B1 patent drawingFigure 2A~2F
  • EP3324310B1 patent drawingFigure 2G~2K

AI summary

An encoder includes a processor, a buffer, and a memory. The memory includes code as instructions that cause the processor to perform a number of steps. The steps include quantizing geometric data associated with a geometric construct, partitioning the geometric construct, determining a number of points in the partition, generating a deviation value based on the number of points in the partition, storing the deviation value in the buffer, and entropy encoding the deviation value.