Multi-Directional Camera Pose Registration via Virtual Image Re-Projection

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing methods for camera pose registration in multi-directional image capture apparatuses, such as 360° camera systems, are inefficient, computationally intensive, and prone to errors, especially when the arrangement of cameras changes over time.

Innovation Solution

The method involves image re-projection to generate virtual camera images based on a previously generated 3D model, determining point correspondences, and using Direct Linear Transform to calculate the positions and orientations of virtual cameras, with outlier filtering to improve accuracy.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If traditional camera pose registration methods are used for multi-directional image capture apparatuses, then the positions and orientations can be determined, but the computational complexity is high and the processing efficiency is low

Engineering Contradiction:
Improvecamera pose registration efficiencyVSAvoidcomputational complexity
Core Design Contradiction:
ProductivityVSDevice complexity

Solution Approach 1:

The patent applies preliminary action by pre-generating a three-dimensional model of the scene before performing camera pose registration. This 3D model is created in advance and stored, so that during registration, the system only needs to compare captured images against the pre-existing model rather than performing full 3D reconstruction in real-time. This significantly reduces the computational complexity and improves processing efficiency when determining camera positions and orientations.

Inventive Principle:
Principle #10Preliminary action

2Reliability

If traditional camera pose registration methods are used, then positions and orientations can be determined, but errors increase when camera arrangements change over time

Engineering Contradiction:
Improveaccuracy of position determinationVSAvoidadaptability to camera arrangement changes
Core Design Contradiction:
ReliabilityVSAdaptability or versatility

Solution Approach 1:

The patent applies parameter changes by using a flexible image matching approach that compares captured images against a pre-generated 3D model without assuming fixed camera positions. The system extracts features from both the captured images and the 3D model, then uses direct linear transform to calculate camera poses based on these feature correspondences. This method adapts to different camera arrangements and temporal changes in the multi-directional capture system, maintaining accuracy even when camera configurations change.

Inventive Principle:
Principle #35Parameter changes

3Measurement precision

If comprehensive processing of all virtual camera images is performed, then accurate position determination is achieved, but the processing time and computational resources increase

Engineering Contradiction:
Improveprecision of camera positionVSAvoidprocessing time
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The patent applies the extraction principle by selecting and using only the essential information needed for camera pose determination. Instead of processing all possible virtual camera images comprehensively, the system extracts key features from captured images and matches them against corresponding features in the pre-generated 3D model. This selective extraction of relevant feature correspondences maintains measurement precision while significantly reducing processing time and computational resource requirements.

Inventive Principle:
Principle #2Taking out (Extraction)

Data Source

PatentUS10565803B2Methods and apparatuses for determining positions of multi-directional image capture apparatuses
Publication Date: 2020.02.18 NOKIA TECHNOLOGIES OY
  • US10565803B2 patent drawing
  • US10565803B2 patent drawing
  • US10565803B2 patent drawing

AI summary

A method comprising performing image re-projection on each of a plurality of first images of a scene, thereby to generate a plurality of re-projected second images of the scene, wherein each first image of the scene is captured by a respective camera of a first multi-directional image capture apparatus and each second image of the scene is associated with a respective virtual camera; processing the plurality of second images based on a previously generated virtual three dimensional model of the scene, thereby to generate respective positions of the virtual cameras associated with the second images; and determining a position of the first multi-directional image capture apparatus based on one or more of the generated positions of the virtual cameras.