Image Registration via Region-Weighted Motion Vectors

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Conventional image registration techniques face challenges in accurately registering images with optical system distortion, particularly when the distortion varies significantly across the image height, making it difficult to perform simultaneous registration at the central and peripheral portions of an image.

Innovation Solution

An image processing apparatus and method that sets multiple motion vector measurement regions, calculates motion vectors within these regions, specifies a region of interest based on user-defined criteria, and adjusts the contribution of each motion vector to prioritize those within the region of interest, thereby integrating motion vectors to determine an inter-image motion vector with enhanced precision.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If block matching method is used to calculate motion vectors across the entire image, then overall image registration can be performed, but registration accuracy deteriorates in regions with large optical distortion

Engineering Contradiction:

Improvemotion vector accuracyVSAvoidregistration reliability

Core Design Contradiction:

Measurement precisionVSReliability

Solution Approach 1:

The image is divided into multiple measurement regions (central region and peripheral regions) with different weights. Motion vectors from the central region are given higher weight than those from peripheral regions, allowing the system to segment the image based on distortion characteristics and prioritize accurate regions in the registration process.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

Different regions of the image are assigned different quality weights based on their distortion characteristics. The central region, which has smaller distortion, is assigned a higher weight (e.g., 0.7) while peripheral regions with larger distortion are assigned lower weights (e.g., 0.3), making the registration process adapt to local quality variations across the image.

Inventive Principle:

Principle #3Local quality

2Area of stationary object

If motion vectors from all regions are equally weighted in integration, then comprehensive image coverage is achieved, but registration precision deteriorates due to inclusion of distorted peripheral regions

Engineering Contradiction:

Improveimage coverage areaVSAvoidregistration precision

Core Design Contradiction:

Area of stationary objectVSMeasurement precision

Solution Approach 1:

The patent applies local quality weighting where the central region (with smaller distortion) is assigned a higher weight and peripheral regions (with larger distortion) are assigned lower weights. This allows comprehensive coverage while maintaining precision by letting high-quality regions contribute more to the final registration result.

Inventive Principle:

Principle #3Local quality

Solution Approach 2:

The patent changes the parameter of motion vector contribution by introducing weight coefficients that vary by region. Instead of equal weighting, the central region's motion vectors are multiplied by a larger weight factor, effectively changing the parameter of how motion vectors are integrated to prioritize accurate regions.

Inventive Principle:

Principle #35Parameter changes

3Measurement precision

If optical system distortion is corrected before motion vector detection, then registration accuracy improves, but processing complexity increases

Engineering Contradiction:

Improveregistration accuracyVSAvoidprocessing complexity

Core Design Contradiction:

Measurement precisionVSDevice complexity

Solution Approach 1:

Instead of performing distortion correction as a preliminary step before motion vector detection, the patent takes the opposite approach by directly detecting motion vectors from the distorted image and then applying region-based weighting to compensate for distortion effects. This avoids the complexity of distortion correction while still achieving accurate registration.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The patent inverts the conventional approach by not correcting distortion before motion vector detection. Instead, it detects motion vectors directly from the distorted image and uses region-based weighting to compensate, effectively solving the problem in reverse to reduce processing complexity.

Inventive Principle:

Principle #13The other way round (Inversion)

Data Source

PatentUS8798130B2Image processing apparatus, electronic device, image processing method, and storage medium storing image processing program

Publication Date: 2014.08.05 OLYMPUS CORPORATION(JP)

AI summary

An image processing apparatus that performs image registration processing between a plurality of images through a motion vector calculation sets a plurality of motion vector measurement regions on an image, calculates a motion vector in each of the plurality of motion vector measurement regions, specifies a region of interest on the image, determines whether or not each of the plurality of motion vector measurement regions is included in the region of interest, calculates a contribution of each motion vector such that the contribution of the motion vector of a motion vector measurement region included in the region of interest is larger than the contribution of the motion vector of a motion vector measurement region not included in the region of interest, and determines an inter-image motion vector by integrating the motion vectors calculated respectively in the plurality of motion vector measurement regions in accordance with the calculated contribution.