Camera Motion Prediction Using Optical Image Stabilization Control

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing image processing techniques fail to accurately predict camera motion in augmented reality applications when using optical image stabilization, leading to inaccuracies in motion synchronization between video and computer graphics, and errors in three-dimensional information acquisition due to increased calculation complexity and instability.

Innovation Solution

An image processing apparatus and method that eliminates the influence of optical image stabilization by using optical image stabilization control information to correct camera motion predictions, allowing for accurate camera work prediction and stabilization, even in conditions with wide-angle lenses and large motion, by transforming corresponding points into normalized image coordinates and applying geometric transformations.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If a complicated camera model with many variables is used to predict camera work by optimization technique, then measurement precision of camera trajectory is improved, but device complexity and calculation amount are exponentially increased

Engineering Contradiction:
Improvecamera trajectory information accuracyVSAvoidcamera model complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent extracts and removes the optical image stabilization effect from the camera work prediction process. By separating the stabilization component from the overall camera motion, the system can use a simpler camera model that doesn't need to account for stabilization mechanics, thereby reducing model complexity while maintaining accurate trajectory prediction through the extracted stabilization information

Inventive Principle:
Principle #2Taking out (Extraction)

Solution Approach 2:

The patent changes the parameter set by using optical image stabilization control information as a separate input parameter rather than incorporating it into a complex camera model. This parameter transformation allows the optimization technique to work with fewer variables while still achieving accurate camera trajectory prediction

Inventive Principle:
Principle #35Parameter changes

2Measurement precision

If a complicated camera model with many variables is used to predict camera work by optimization technique, then measurement precision of camera trajectory is improved, but calculation convergence and processing stability deteriorate

Engineering Contradiction:
Improvecamera trajectory information accuracyVSAvoidprocessing stability
Core Design Contradiction:
Measurement precisionVSReliability

Solution Approach 1:

By extracting the optical image stabilization effect as a separate component, the patent removes a source of computational complexity and instability from the optimization process. This separation allows the optimization algorithm to converge more reliably while maintaining accurate trajectory prediction

Inventive Principle:
Principle #2Taking out (Extraction)

Solution Approach 2:

The optical image stabilization control information serves as an intermediary that bridges the gap between the simple camera model and the need for accurate trajectory prediction. This intermediary provides the necessary stabilization data without requiring the system to use a complex camera model, thereby maintaining processing stability

Inventive Principle:
Principle #24Intermediary (Mediator)

3Manufacturing precision

If optical image stabilization mechanism is introduced to suppress blur, then image quality is improved, but motion shift between shot video and CG stands out

Engineering Contradiction:
Improveimage qualityVSAvoidcamera work information accuracy
Core Design Contradiction:
Manufacturing precisionVSMeasurement precision

Solution Approach 1:

The patent converts the harmful effect of motion shift caused by optical image stabilization into a beneficial solution by extracting and utilizing the stabilization control information. The same stabilization mechanism that causes motion shift is now used to correct the prediction, turning the problem into a solution that improves both image quality and camera work accuracy

Inventive Principle:
Principle #22Blessing in disguise (Convert harm into benefit)

Solution Approach 2:

The system uses optical image stabilization control information as feedback to correct camera work predictions. By feeding back the stabilization data into the prediction process, the system compensates for the motion shift introduced by stabilization, maintaining accuracy in AR applications and three-dimensional information acquisition

Inventive Principle:
Principle #23Feedback

Data Source

PatentEP2785045B1Image processing apparatus and method, and image capturing apparatus
Publication Date: 2017.12.27 CANON KK
  • EP2785045B1 patent drawingFigure 1
  • EP2785045B1 patent drawingFigure 2~3
  • EP2785045B1 patent drawingFigure 4

AI summary

An image processing apparatus comprises: input means for inputting an image signal obtained by photoelectrically converting, by an image sensor, an optical image incident via an optical system including optical image stabilization means; acquisition means for acquiring optical image stabilization control information obtained by the optical image stabilization means; and motion of camera prediction means for predicting a motion of camera based on information obtained by eliminating influence of optical image stabilization by the optical image stabilization means from the image signal input from the input means based on the optical image stabilization control information.