Visual Image Pattern Matching Aperture Displacement Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing visual image processing techniques fail to accurately detect time-based movement in images due to aperture problems, where selected features become geometrically indistinct from neighboring regions, and are sensitive to noise, leading to incorrect motion estimation.
Innovation Solution
A method is developed to detect time-elapsed movement by selecting a reference region and a new location in a visual image, forming a motion estimation block, and creating a distance map with motion estimation vectors, while incorporating symmetry detection and noise reduction algorithms to improve accuracy and confidence in motion estimation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If pattern matching algorithms are used to detect motion in visual images, then motion estimation can be performed, but accuracy deteriorates when aperture problems occur where features are geometrically indistinct from neighboring regions
Solution Approach 1:
The patent applies asymmetry by detecting symmetry properties of image features and using asymmetric treatments for symmetric versus asymmetric regions. Specifically, the system identifies whether a feature exhibits symmetry (e.g., horizontal or vertical symmetry) and applies different motion estimation strategies: for symmetric features, it searches along the symmetry axis only, while for asymmetric features, it uses conventional 2D search. This resolves the aperture problem by exploiting the inherent asymmetry in most natural features to disambiguate motion direction.
Solution Approach 2:
The patent segments the motion estimation process into distinct phases: symmetry detection, symmetry axis identification, and conditional motion search. By dividing the problem into these segments, the system can apply specialized algorithms to each phase. The segmentation allows the system to first identify symmetric features, then determine their symmetry axes, and finally perform motion estimation only along relevant directions, improving both accuracy and efficiency.
2Productivity
If pattern matching techniques are used to track object displacement, then motion estimation is achieved, but noise sensitivity increases leading to incorrect similarity measures
Solution Approach 1:
The patent applies local quality by adapting the motion estimation strategy to local feature properties. Instead of using a uniform approach across the entire image, the system analyzes each feature's symmetry characteristics locally and applies customized search strategies. For example, horizontally symmetric features trigger a vertical-only search, while vertically symmetric features trigger a horizontal-only search. This local adaptation reduces noise sensitivity by constraining the search space to directions where motion is actually expected.
Solution Approach 2:
The patent introduces symmetry detection as an intermediary step between traditional pattern matching and motion estimation. This intermediary process analyzes the geometric properties of features and uses this information to guide the subsequent motion search. The symmetry analysis acts as a mediator that filters out noise-induced false matches by pre-establishing constraints based on geometric invariance, thereby improving the reliability of similarity measures.
3Speed
If conventional motion estimation algorithms are used, then processing speed is maintained, but accuracy deteriorates for features with vertical and horizontal displacement information
Solution Approach 1:
The patent applies dynamics by making the search strategy adaptive rather than static. The system dynamically adjusts the search dimensions based on the detected symmetry properties of each feature. For symmetric features, the search is constrained to one dimension (along the symmetry axis), while for asymmetric features, the full 2D search is performed. This dynamic adaptation allows the system to maintain high processing speed for symmetric features while ensuring accurate displacement detection for all feature types.
Solution Approach 2:
The patent changes the search parameter space based on feature symmetry. Instead of always searching both horizontal and vertical dimensions, the system modifies the search parameters dynamically: for horizontally symmetric features, it sets vertical search range to zero; for vertically symmetric features, it sets horizontal search range to zero. This parameter change reduces the computational complexity from O(W×H) to O(max(W,H)) for symmetric features, thereby maintaining processing speed while improving accuracy.
Data Source
AI summary
Exemplary embodiments of the invention as described herein generally provide for detecting the displacement of feature(s) within a visual image in cases where pattern matching fails due to the existence of aperture(s) caused for example by external condition(s) encountered in recording such an image over time. Technique(s) are disclosed for detecting the difference between displacement of a geometric feature of an object appearing within an image (e.g., an edge or smooth surface) that has an aperture and another feature (e.g., a corner) that does not since it is not symmetrically invariant.