Geometry Encoding Parameter Selection Using Data Signatures

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Selecting optimal compression techniques and parameters for geometric data is challenging due to the variety and complexity of available models, making it difficult for users to achieve efficient compression results.

Innovation Solution

A machine learning-based system that selects the best compression technique and parameters for geometric data by using a trained model to determine a unique signature of the data and iteratively testing different options, combining fixed and environmental parameters with machine learning algorithms to optimize encoding and decoding performance.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Manufacturing precision

If the number and variety of compression models increase, then compression quality may improve, but the difficulty of selecting optimal models and parameters increases

Engineering Contradiction:
Improvecompression qualityVSAvoidmodel selection complexity
Core Design Contradiction:
Manufacturing precisionVSDevice complexity

Solution Approach 1:

The system automatically selects compression models and parameters by analyzing geometric data properties (vertex valence, angle histograms) and querying a pre-built index of compression models, eliminating the need for user manual selection and reducing selection complexity while maintaining high compression quality

Inventive Principle:
Principle #25Self-service

Solution Approach 2:

The system changes the approach from manual parameter selection to automated parameter determination by using machine learning models that map geometric data characteristics to optimal compression parameters, thereby reducing complexity while improving compression quality

Inventive Principle:
Principle #35Parameter changes

2Manufacturing precision

If the number of parameters associated with compression models increases, then compression optimization may improve, but the difficulty of selecting desirable parameters increases

Engineering Contradiction:
Improvecompression optimizationVSAvoidparameter selection ease
Core Design Contradiction:
Manufacturing precisionVSEase of operation

Solution Approach 1:

The system automatically determines optimal parameter values by analyzing geometric data properties and querying the compression model index, eliminating the need for users to manually adjust multiple parameters while achieving optimized compression results

Inventive Principle:
Principle #25Self-service

Solution Approach 2:

The system introduces an intermediary layer (machine learning model and index structure) that automatically maps geometric data characteristics to optimal compression parameters, shielding users from parameter selection complexity while maintaining optimization performance

Inventive Principle:
Principle #24Intermediary (Mediator)

3Productivity

If compression models become more complex, then compression performance may improve, but the computational complexity increases

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

Solution Approach 1:

The system segments the compression process into two independent phases: a training phase where complex compression models are pre-built and indexed, and an inference phase where simple data analysis and index querying are performed, thereby reducing online computational complexity while maintaining compression performance

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The system performs preliminary action by pre-building and indexing compression models during training, so that during actual compression operations, the system only needs to query pre-computed models rather than perform complex computations in real-time, reducing computational complexity while maintaining performance

Inventive Principle:
Principle #10Preliminary action

Data Source

PatentEP3803794B1Geometry encoder
Publication Date: 2026.01.28 GOOGLE LLC
  • EP3803794B1 patent drawingFigure 1
  • EP3803794B1 patent drawingFigure 2
  • EP3803794B1 patent drawingFigure 3

AI summary

A method includes receiving geometric data to be encoded, generating a signature for the geometric data based on the at least one property associated with the geometric data, enumerating a first set of options, enumerating a second set of options, encoding the geometric data using the first option and the second option, decoding the encoded geometric data, determining a performance associated with encoding the geometric data, determining a performance associated with decoding the encoded geometric data, and training a regressor based on the signature, the enumerated first option, the enumerated second option, the performance associated with encoding the geometric data and the performance associated with decoding the encoded geometric data.