Image Matching Using Topological Equivalence and Orientation Tests
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image matching algorithms, such as RANSAC, often generate physically invalid transformations and fail to ensure rotational invariance, leading to incorrect matching and unnecessary computation, especially when images share common elements like logos or text, and may not accurately determine matches between query and database images.
Innovation Solution
A method that performs a topological equivalence test and orientation consistency test before determining transformations, followed by a coverage test to ensure valid matches, using putative correspondences and inliers to identify final matches based on predefined thresholds for the number of inliers and Root Mean Square error.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If RANSAC algorithm is used to generate transformations from putative correspondences, then the matching process can be performed, but physically invalid transformations (such as flipping or twisting) may be generated leading to incorrect matching
Solution Approach 1:
The patent applies preliminary action by performing a topological equivalence test on putative correspondences before generating transformations. This pre-validation step checks whether the correspondence set could possibly produce a physically valid transformation, eliminating invalid candidates before the actual transformation generation occurs, thus preventing wasted computation and incorrect matches
Solution Approach 2:
The patent introduces an intermediary validation mechanism (topological equivalence test) that acts as a mediator between putative correspondences and transformation generation. This intermediary layer filters out correspondences that would lead to physically invalid transformations, ensuring only valid candidates proceed to transformation generation
2Productivity
If statistical algorithms like RANSAC are used to process putative correspondences, then geometric matching can be achieved, but useless computation occurs on physically invalid transformations
Solution Approach 1:
The patent performs preliminary topological equivalence testing on putative correspondences before they are processed by RANSAC. This preliminary action identifies and eliminates correspondence sets that would inevitably produce invalid transformations, preventing wasted computational resources on futile processing operations
Solution Approach 2:
The patent applies partial action by selectively processing only those putative correspondences that pass the topological equivalence test. Rather than processing all putative correspondences uniformly, it applies the computationally intensive RANSAC algorithm only to the valid subset, reducing overall computation waste while maintaining matching effectiveness
3Productivity
If feature points are matched based on visual similarity alone, then putative correspondences can be found quickly, but geometric consistency cannot be ensured
Solution Approach 1:
The patent segments the matching process into distinct stages: first identifying putative correspondences based on visual similarity (fast but imprecise), then validating them through topological equivalence testing (geometric consistency check). This segmentation allows the system to benefit from both rapid initial detection and rigorous geometric verification
Solution Approach 2:
The patent performs preliminary visual matching to generate putative correspondences quickly, then applies preliminary geometric validation through topological equivalence testing. This two-stage preliminary approach ensures both speed in correspondence generation and accuracy in geometric consistency verification before final transformation generation
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A method, system and computer program product for matching images is provided. The images to be matched are represented by feature points and feature vectors and orientations associated with the feature points. First, putative correspondences are determined by using feature vectors. A subset of putative correspondences is selected and the topological equivalence of the subset is determined. The topologically equivalent subset of putative correspondences is used to establish a motion estimation model. An orientation consistency test is performed on the putative correspondences and the corresponding motion estimation transformation that is determined, to avoid an infeasible transformation. A coverage test is performed on the matches that satisfy orientation consistency test. The candidate matches that do not cover a significant portion of one of the images are rejected. The final match images are provided in the order of decreasing matching, in case of multiple images satisfying all the test requirements.