3D Modeling from Virtual Viewpoint Images

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing techniques for modeling three-dimensional objects from moving images are limited, as they can only select objects from pre-defined highlight scenes, making it difficult to model objects not included in these scenes.

Innovation Solution

An information processing apparatus with an identification unit, an obtaining unit, and an output unit that generates modeling data for three-dimensional objects based on shape data from virtual viewpoint images, allowing users to specify targets and generate data for objects in any scene by adjusting virtual camera positions and angles.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Ease of operation

If object selection is limited to pre-defined highlight scenes, then the system complexity is reduced and operation is simplified, but the adaptability and versatility of the modeling system deteriorates

Engineering Contradiction:
Improveobject selection processVSAvoidscene coverage capability
Core Design Contradiction:
Ease of operationVSAdaptability or versatility

Solution Approach 1:

The system enables a single modeling apparatus to handle multiple functions: it can process both pre-defined highlight scenes and arbitrary user-specified scenes within the same moving image data, making the system universally applicable to diverse modeling needs without requiring separate systems

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

2Adaptability or versatility

If the system processes arbitrary scenes instead of only pre-defined highlight scenes, then the adaptability and versatility improve, but the device complexity and operational difficulty increase

Engineering Contradiction:
Improvescene selection flexibilityVSAvoidprocessing system complexity
Core Design Contradiction:
Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The system allows users to independently specify arbitrary scenes and objects within moving image data using intuitive operations on visualized time information and virtual viewpoint images, enabling self-service modeling without requiring complex system configuration or expert intervention

Inventive Principle:
Principle #25Self-service

3Loss of time

If pre-defined highlight scenes are used, then the ease of operation is improved and processing time is reduced, but the measurement precision and modeling accuracy for desired objects deteriorates

Engineering Contradiction:
Improvescene selection timeVSAvoidobject identification accuracy
Core Design Contradiction:
Loss of timeVSMeasurement precision

Solution Approach 1:

The system generates virtual viewpoint images and visualizes time information to provide feedback to users about available scenes and objects, allowing users to verify and adjust their selections before finalizing the modeling process, thereby ensuring high precision in object identification

Inventive Principle:
Principle #23Feedback

Data Source

PatentUS12062137B2Information processing apparatus, information processing method, and storage medium
Publication Date: 2024.08.13 CANON KK
  • US12062137B2 patent drawing
  • US12062137B2 patent drawing
  • US12062137B2 patent drawing

AI summary

An information processing apparatus has an identification unit, an obtaining unit, and an output unit. The identification unit identifies, in a virtual viewpoint image generated using shape data representing a three-dimensional shape of an object, a target for which to generate data for modeling a three-dimensional object, based on time information related to the virtual viewpoint image as well as a position of a virtual viewpoint and a direction of a line of sight from the virtual viewpoint related to the virtual viewpoint image. The obtaining unit obtains shape data on a first object in the virtual viewpoint image, the first object corresponding to the identified target. The output unit outputs data for modeling a three-dimensional object generated based on the obtained shape data.