Image Registration Using Spatial Region Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image registration techniques, such as those used in non-rigid registration, fail to accurately evaluate similarity between images based on spatial characteristics, leading to erroneous judgments when multiple subjects with similar pixel values but different spatial positions are compared.
Innovation Solution
An image processing apparatus and method that divides images into regions based on predetermined conditions, such as distance or angle from reference points, and evaluates similarity using an evaluation function that considers the correlation between pixel value distributions in corresponding regions, thereby accurately reflecting spatial features in similarity assessments.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If the degree of similarity between two images is judged based only on the correlative properties between the distributions of pixel values, then the evaluation process is simple, but the judgment of similarity becomes erroneous when multiple subjects with similar pixel values but different spatial positions are compared
Solution Approach 1:
The patent divides the image into multiple regions based on spatial characteristics (e.g., distance from reference points, angular ranges) and evaluates the correlation between pixel value distributions within each corresponding region. This segmentation approach allows the system to maintain computational simplicity while improving similarity judgment accuracy by considering both spatial position and pixel value distribution together rather than pixel values alone.
2Measurement precision
If images are divided into regions based on spatial characteristics and similarity is evaluated using correlation between pixel value distributions in corresponding regions, then the judgment accuracy improves, but the computational complexity increases
Solution Approach 1:
The patent applies different evaluation approaches to different regions of the image. By dividing the image into multiple regions based on spatial characteristics and evaluating the correlation between pixel value distributions within each corresponding region, the system achieves accurate similarity judgment while maintaining manageable computational complexity through localized analysis rather than global processing.
3Ease of manufacture
If the total number of pixels belonging to the same range of pixel values are the same within two images, then the conventional method judges the images as similar, but this leads to erroneous judgment when the numbers or spatial positions of subjects are different
Solution Approach 1:
The patent adds a spatial dimension to the similarity evaluation by dividing images into regions based on spatial characteristics (distance from reference points, angular ranges) and evaluating correlation within each region. This transforms the evaluation from a single-dimensional pixel value comparison to a multi-dimensional assessment that incorporates both spatial position and pixel value distribution, thereby improving reliability while maintaining simplicity.
Data Source
AI summary
A first image and a second image are obtained; the amount of deformation of the first image is estimated by evaluating the degree of similarity between a deformed first image and the second image, using an evaluation function that evaluates the correlation between the distribution of corresponding pixel values within the two images; and an image, which is the first image deformed based on the estimated amount of deformation, is generated. The evaluation function evaluates the degree of similarity between the deformed first image and the second image, based on degrees of similarities of divided images that represent degrees of similarities among the distributions of pixel values of each pair of divided first images and divided second images, which respectively are images that the deformed first image is divided into and images that the second image is divided into, according to predetermined dividing conditions.