Feature Point Matching Using Position and Pixel Value Features
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing feature point matching techniques struggle to accurately match feature points from images that have undergone geometric transformations such as affine and projective transformations, as they rely solely on feature point positions, leading to accuracy deterioration.
Innovation Solution
A feature point matching device and method that generates position and pixel value feature quantities for each feature point, allowing for matching based on the degree of similarity between these quantities across images, even under geometric transformations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If feature point matching is performed using only position coordinates, then the matching process is simple and fast, but matching accuracy deteriorates when images undergo geometric transformations
Solution Approach 1:
The patent combines position feature quantities (spatial coordinates) and pixel value feature quantities (image content) to form a comprehensive matching criterion. This merging allows the system to maintain fast processing while achieving high accuracy under geometric transformations, as both positional and appearance information are utilized simultaneously in the matching process
Solution Approach 2:
The patent introduces multiple feature quantity parameters (position and pixel values) instead of relying solely on position coordinates. By changing the parameter set used for matching, the system achieves robustness against geometric transformations while maintaining computational efficiency through the structured approach of calculating and comparing these dual parameters
2Device complexity
If feature point matching uses only position information, then the computational complexity is low, but reliability of matching deteriorates under affine and projective transformations
Solution Approach 1:
The patent merges position feature quantities and pixel value feature quantities into a unified matching framework. This combination enhances matching reliability under affine and projective transformations by utilizing both spatial and appearance information, while the systematic approach to calculating these features keeps computational complexity manageable
3Measurement precision
If both position and pixel value features are used for matching, then matching accuracy improves under geometric transformations, but the complexity of feature generation and processing increases
Solution Approach 1:
The patent segments the feature generation process into distinct modules: position feature quantity calculation and pixel value feature quantity calculation. This segmentation allows each module to be optimized independently, managing complexity while achieving high matching accuracy through the combination of both feature types
Data Source
AI summary
A feature point matching device includes a position feature generation unit for generating a position feature quantity for a feature point based on a position of the feature point included in an image, a pixel value feature generation unit for generating a pixel value feature quantity for the feature point based on a pixel value of the feature point included in the image, and a feature matching unit for matching a first feature point and a second feature point based on a degree of matching between a first position feature quantity generated from a first image and a second position feature quantity generated from a second image and a degree of matching between a first pixel value feature quantity generated from the first image and a second pixel value feature quantity generated from the second image.