Template Creation Device Clustering and Integration for 3D Recognition

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current template matching techniques for 3D object recognition are inefficient due to the high number of templates required, especially for symmetrical or simply shaped objects, leading to increased processing time and memory costs, as they fail to effectively reduce the number of templates and combine similar viewpoints.

Innovation Solution

A template creation device that integrates and clusters templates with similar image features by lowering resolution, creating multiple layers of template sets for efficient recognition, using low-resolution templates for rough recognition and high-resolution templates for detailed recognition, and computing similarity scores based on feature points to reduce processing time.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If individual templates are prepared for each pose to enable 3D object recognition through template matching, then recognition accuracy is improved, but the number of templates increases significantly leading to increased processing time

Engineering Contradiction:
Improverecognition accuracyVSAvoidprocessing time
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The patent merges multiple templates representing different poses of the same object into a single integrated template. The integration unit combines templates that have high similarity scores (above a threshold) into unified templates, reducing the total number of templates while preserving recognition accuracy across multiple poses.

Inventive Principle:
Principle #5Merging (Combining)

Solution Approach 2:

The integrated templates serve multiple functions by representing multiple poses simultaneously. A single integrated template can match objects in various poses, making the template set more universal and reducing the need for pose-specific templates.

Inventive Principle:
Principle #6Universality (Multi-functionality)

2Loss of time

If the number of viewpoints is reduced to decrease template count, then processing time is reduced, but the technique cannot sufficiently reduce template matching iterations for symmetrical or simply shaped objects

Engineering Contradiction:
Improveprocessing timeVSAvoidrecognition accuracy
Core Design Contradiction:
Loss of timeVSMeasurement precision

Solution Approach 1:

The patent changes the parameter used for template grouping from viewpoint distance to image feature similarity. By using similarity scores based on image features rather than geometric viewpoint distances, the system can effectively group templates for symmetrical and simply shaped objects that appear similar from different viewpoints.

Inventive Principle:
Principle #35Parameter changes

3Measurement precision

If high-resolution templates are used for all recognition tasks, then recognition accuracy is maintained, but memory requirements and processing load increase

Engineering Contradiction:
Improverecognition accuracyVSAvoidmemory requirements
Core Design Contradiction:
Measurement precisionVSQuantity of substance

Solution Approach 1:

The patent applies different resolutions to different templates based on their specific needs. Not all templates are stored or processed at high resolution - the system uses resolution appropriate to each template's contribution to recognition accuracy, optimizing the balance between quality and resource usage.

Inventive Principle:
Principle #3Local quality

Data Source

PatentEP3258441B1Template creation device and template creation method
Publication Date: 2022.06.08 OMRON CORP
  • EP3258441B1 patent drawingFigure 1
  • EP3258441B1 patent drawingFigure 2
  • EP3258441B1 patent drawingFigure 3

AI summary

To reduce the number of templates and effectively combine the templates used for recognizing an object, a template creation device includes an acquisition unit configured to acquire a plurality of templates from a plurality of images of different poses of a single object, or a plurality of images for a plurality of objects; a clustering unit configured to divide the plurality of templates into a plurality of groups on the basis of a similarity score; and an integration unit configured to combine the templates in a group into an integrated template, and to create a new template set from the plurality of integrated templates corresponding to each group in the plurality of groups.