Template Creation Device Clustering and Integration for 3D Recognition
Find Innovative SolutionsGenerate 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
Engineering 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
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.
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.
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
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.
3Measurement precision
If high-resolution templates are used for all recognition tasks, then recognition accuracy is maintained, but memory requirements and processing load increase
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.
Data Source
Figure 1
Figure 2
Figure 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.