Structured Light Pattern Codeword Arrangement for 3D Reconstruction Error Correction

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing 3D reconstruction systems using structured light patterns face errors due to misrecognition of projected patterns, particularly in environments with static codeword orders and specific a-priori conditions, which limits their error-correcting capabilities.

Innovation Solution

The system employs a pattern projection method where symbols (codewords) are arranged such that each symbol is more dissimilar to its proximal neighbors than to its distal neighbors, ensuring a larger Hamming distance for proximal symbol pairs, thereby enhancing error-correcting capabilities within a local neighborhood, and dynamically configures the pattern projector to illuminate specific ranges, ensuring accurate 3D location computation.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If traditional structured light patterns with static codeword orders are used, then the system structure is simple, but error-correcting capabilities are limited

Engineering Contradiction:
Improveerror-correcting capabilityVSAvoidpattern arrangement complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The patent applies local quality by making each local neighborhood of symbols have unique dissimilarity properties. Specifically, each symbol is arranged to have maximum dissimilarity from its proximal neighbors while being less dissimilar from distal neighbors, creating localized error-correcting zones that improve overall pattern recognition reliability without requiring complete system redesign

Inventive Principle:
Principle #3Local quality

Solution Approach 2:

The patent employs asymmetry in the symbol arrangement where the dissimilarity relationship is not uniform across all symbol pairs. Instead, each symbol has an asymmetric dissimilarity profile relative to its neighbors at different distances, with proximal neighbors being maximally dissimilar and distal neighbors being less dissimilar, thereby enhancing local error detection and correction capabilities

Inventive Principle:
Principle #4Asymmetry

2Measurement precision

If symbols are arranged with high dissimilarity to proximal neighbors, then error correction improves, but pattern complexity increases

Engineering Contradiction:
Improve3D location accuracyVSAvoidsymbol arrangement complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent implements local quality by optimizing the dissimilarity properties specifically for local neighborhoods of symbols. Each symbol's immediate neighbors are made maximally dissimilar to enable local error correction, while maintaining a systematic pattern that doesn't require complete global redesign, thus improving measurement precision with controlled complexity increase

Inventive Principle:
Principle #3Local quality

Solution Approach 2:

The patent segments the overall pattern into local neighborhoods where each symbol's relationship with its proximal neighbors is independently optimized for maximum dissimilarity. This segmentation allows error correction to be performed locally without requiring complex global pattern analysis, improving 3D location accuracy while managing overall pattern complexity

Inventive Principle:
Principle #1Segmentation

Data Source

PatentEP3444782B1Coding distance topologies for structured light patterns for 3D reconstruction
Publication Date: 2022.03.30 COGNEX CORP
  • EP3444782B1 patent drawingFigure 1A
  • EP3444782B1 patent drawingFigure 1B
  • EP3444782B1 patent drawingFigure 2

AI summary

Methods, systems, and devices for 3D measurement and/or pattern generation are provided in accordance with various embodiments. Some embodiments include a method of pattern projection that may include projecting one or more patterns. Each pattern from the one or more patterns may include an arrangement of three or more symbols that are arranged such that for each symbol in the arrangement, a degree of similarity between said symbol and a most proximal of the remaining symbols in the arrangement is less than a degree of similarity between said symbol and a most distal of the remaining symbols in the arrangement. Some embodiments further include: illuminating an object using the one or more projected patterns; collecting one or more images of the illuminated object; and/or computing one or more 3D locations of the illuminated object based on the one or more projected patterns and the one or more collected images.