Image Feature Point Thinning Through Hierarchical Region Merging
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing technologies struggle to efficiently adjust the number of feature points across an entire image, making it difficult to perform thinning-out processes effectively.
Innovation Solution
An image processing device that divides an image into small regions, generates combination regions by combining adjacent small regions, and thins out feature points based on their scores in descending order until a predetermined termination condition is met, allowing for efficient adjustment of feature points across the entire image.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If the image is divided into a plurality of regions and thinning-out is performed in each divided region, then the thinning-out process can be performed locally in each region, but the number of feature points in the entire image cannot be efficiently adjusted
Solution Approach 1:
The patent combines multiple small regions into combination regions (e.g., 2x2 or 3x3 grids of small regions) and performs thinning-out at the combination region level. This merging approach enables efficient control of the total number of feature points across the entire image while maintaining the ability to perform localized thinning-out. The combination regions serve as intermediate units that bridge local and global control requirements.
2Ease of operation
If feature points are thinned out in ascending order of reliability in each divided region, then local thinning-out is achieved, but uniform distribution of feature points across the entire image cannot be ensured
Solution Approach 1:
By merging small regions into combination regions and performing thinning-out at the combination region level, the patent ensures uniform feature point distribution across the entire image. The combination regions provide a larger context that enables balanced distribution while preserving local thinning-out capabilities through the hierarchical structure.
Solution Approach 2:
The patent introduces a hierarchical dimension by creating multiple levels of regions (small regions within combination regions). This dimensional hierarchy enables simultaneous achievement of local thinning-out (at small region level) and uniform global distribution (at combination region level), resolving the contradiction between local control and global uniformity.
3Productivity
If multiple combination regions are generated by combining adjacent small regions, then efficient adjustment of feature points across the entire image is enabled, but the device complexity increases
Solution Approach 1:
The patent segments the image processing into hierarchical levels (small regions and combination regions), which simplifies the management complexity. By dividing the large-scale problem into smaller, manageable units with clear hierarchical relationships, the system can efficiently adjust feature points across the entire image without overwhelming complexity.
Solution Approach 2:
The patent employs dynamic region combining where small regions are flexibly grouped into combination regions based on processing requirements. This dynamic approach allows the system to adapt the region structure to optimize feature point adjustment efficiency while managing complexity through standardized combining rules.
Data Source
AI summary
Multiple feature points are detected from an image. A feature point score is calculated for each feature point. The feature points are thinned out using the feature point scores. In the thinning-out, (a) the image is divided into small regions. (b) Multiple combination regions are generated by combining the small regions adjacent to each other. (c) From among the feature points disposed in each combination region, feature points in descending order of feature point score are selected, and the other feature points are deleted from each combination region. A thinning-out step of processes of (b) and (c) with the combination regions as new small regions is repeated until a predetermined termination condition is met.