3D Polygon Model Vertex Integration for Stereoscopic Image Shape Replication

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

The existing technologies for generating stereoscopic images by integrating multiple 3D polygon models often result in a low degree of replication of the object's shape.

Innovation Solution

An information processing apparatus that acquires distance images from multiple sensors, generates 3D polygon models, divides a virtual space, compares vertex counts, selects vertices with the highest counts, and integrates them to generate a stereoscopic image.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Quantity of substance

If multiple 3D polygon models are integrated to generate a stereoscopic image, then the completeness of object coverage is improved, but the degree of replication of the object's shape deteriorates

Engineering Contradiction:
Improvecompleteness of object coverageVSAvoiddegree of replication of object shape
Core Design Contradiction:
Quantity of substanceVSManufacturing precision

Solution Approach 1:

The integration process is segmented by dividing the virtual space into multiple divided spaces and processing each space independently. This allows selective vertex integration based on local vertex density in each divided space, maintaining high shape replication accuracy while achieving complete object coverage across multiple 3D polygon models.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent applies different integration strategies to different regions of the object by evaluating vertex density locally in each divided space. Vertices are selectively integrated based on their local density characteristics, ensuring that regions with higher vertex density contribute more to the final stereoscopic image, thus improving overall shape replication accuracy.

Inventive Principle:
Principle #3Local quality

2Ease of manufacture

If vertices from all 3D polygon models are integrated equally, then the processing simplicity is maintained, but the accuracy of shape representation deteriorates

Engineering Contradiction:
Improveprocessing simplicityVSAvoidaccuracy of shape representation
Core Design Contradiction:
Ease of manufactureVSManufacturing precision

Solution Approach 1:

The patent changes the integration parameter from equal-weight integration to density-based selective integration. By calculating vertex density in each divided space and using it as a selection criterion, the system achieves accurate shape representation while maintaining relatively simple processing through automated density-based selection rules.

Inventive Principle:
Principle #35Parameter changes

3Manufacturing precision

If vertex selection is based on highest density in each divided space, then the shape replication accuracy is improved, but the computational complexity increases

Engineering Contradiction:
Improveshape replication accuracyVSAvoidcomputational complexity
Core Design Contradiction:
Manufacturing precisionVSDevice complexity

Solution Approach 1:

The computational complexity is managed by segmenting the virtual space into divided spaces and performing vertex density calculations independently in each space. This segmentation reduces the overall computational burden compared to analyzing the entire virtual space at once, while still achieving accurate shape replication through localized high-density vertex selection.

Inventive Principle:
Principle #1Segmentation

Data Source

PatentUS12315082B2Information processing apparatus, non-transitory computer readable medium, and information processing method
Publication Date: 2025.05.27 TOYOTA JIDOSHA KK
  • US12315082B2 patent drawing
  • US12315082B2 patent drawing
  • US12315082B2 patent drawing

AI summary

An information processing apparatus generates a stereoscopic image represented by a collection of polygons. The information processing apparatus includes a controller configured to acquire a plurality of distance images from a plurality of distance image sensors that capture an object, generate a plurality of 3D polygon models based on the plurality of distance images, divide a virtual space surrounding the object to generate a plurality of divided spaces, compare a number of vertices of polygons in the plurality of 3D polygon models in each divided space, select vertices of the 3D polygon model with a highest number of vertices in each divided space as vertices in the divided space, and integrate the selected vertices to generate the stereoscopic image.