IMU-Aided Image Registration for Visual Tracking

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Image registration systems for visual object tracking face challenges when no common landmarks are available or when repetitive patterns cause ambiguous feature correspondence, leading to decreased performance in tracking and guidance systems.

Innovation Solution

An image registration system that utilizes inertial measurement data from an IMU to register image data, enabling alignment even in non-stationary or non-descriptive background scenes with ambiguous features, by applying translation and rotation operations based on sensed motion, and employing hybrid and intensity-based registration modules to identify and align image features.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If image registration is performed via stationary landmarks, then registration accuracy is improved, but performance decreases when no common landmarks are available or when repetitive landmarks cause ambiguous feature correspondence

Engineering Contradiction:
Improveregistration accuracyVSAvoidtracking performance
Core Design Contradiction:
Measurement precisionVSReliability

Solution Approach 1:

The patent introduces an inertial measurement unit (IMU) as an intermediary device that provides motion data to assist image registration. The IMU acts as a mediator between the image registration system and the moving body, providing reliable motion information when visual landmarks are unavailable or ambiguous, thereby resolving the contradiction between registration accuracy and tracking reliability

Inventive Principle:
Principle #24Intermediary (Mediator)

Solution Approach 2:

The system dynamically changes registration parameters by switching between landmark-based registration and IMU-based registration depending on the availability and reliability of landmarks. When landmarks are unavailable or ambiguous, the system transitions to using IMU motion data, changing the fundamental parameter basis for registration from visual features to inertial measurements

Inventive Principle:
Principle #35Parameter changes

2Device complexity

If image feature selection alone is used for registration, then system complexity is reduced, but registration performance deteriorates in areas devoid of landmarks or with ambiguous features

Engineering Contradiction:
Improvesystem complexityVSAvoidregistration performance
Core Design Contradiction:
Device complexityVSMeasurement precision

Solution Approach 1:

The patent implements a dynamic registration system that adapts its complexity based on environmental conditions. The system dynamically switches between simple feature-based registration and more complex hybrid registration involving IMU data, adjusting the level of system complexity according to the availability of reliable landmarks and the specific operational context

Inventive Principle:
Principle #15Dynamics

Data Source

PatentEP3361446B1IMU-aided image registration
Publication Date: 2020.01.08 ROSEMOUNT AEROSPACE INC
  • EP3361446B1 patent drawingFigure 1
  • EP3361446B1 patent drawingFigure 2
  • EP3361446B1 patent drawingFigure 3

AI summary

An imager device (18) disposed on a moving body captures first image data at a first time and second image data at a second, subsequent time. An inertial measurement unit (IMU) disposed on the moving body senses motion of the moving body between the first time and the second time. The first image data is registered to the second image data based on inertial measurement data corresponding to the sensed motion to produce first registered image data. In response to identifying that image features are common to both the first registered image data and the second image data, the first registered image data is registered to the second image data based on the identified common features to produce output registered image data. In response to determining that no image features are common to both the first registered image data and the second image data, the first registered image data is output.