.png){kind=icon}
.png){kind=icon}
Return
Density-Guided Locality Consensus for Fast Feature Matching
.png){kind=icon}
PRE
AI
DOI：10.1109/LGRS.2022.3143102.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Establishing correct feature correspondences from two images of the same scene or target is a critical prerequisite in various applications. This letter proposes a simple but novel method to efficiently remove outliers from a putative match set, which is performed on searching potential inliers that can well preserve local topology structure. To this end, we provide a mathematical formulation and its closed-form solution for fast optimization. To construct local structure more accurately, we introduce a density-guided strategy, which enables our method to distinguish inliers and outliers more easily, thus largely enhancing the matching performance. Extensive experiments on both rigid and nonrigid datasets demonstrate the superiority of our method against the state of the art in both accuracy and efficiency.
Keywords:
Strain
Task analysis
Costs
Feature extraction
Cost function
Topology
Real-time systems
Density-guided
feature matching
locality consensus
outlier
Journal
IF:
16.4
Papers:
1.0W
Citations:
5.1K