THE CORRESPONDING POINTS SCREENING ALGORITHM BASED ON GAUSSIAN KERNEL FUZZY CLUSTERING
Corresponding points matching is the basis of three-dimensional reconstruction, but mismatching often occurs in feature matching. Existing algorithms for handling mismatches, such as RANSAC, mostly use the distance from the point to the polar line (i.e., the residual) to determine whether the matchi...
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Copernicus Publications
2019-09-01
|
| Series: | The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences |
| Online Access: | https://www.int-arch-photogramm-remote-sens-spatial-inf-sci.net/XLII-2-W16/215/2019/isprs-archives-XLII-2-W16-215-2019.pdf |
| Summary: | Corresponding points matching is the basis of three-dimensional reconstruction, but mismatching often occurs in feature matching. Existing algorithms for handling mismatches, such as RANSAC, mostly use the distance from the point to the polar line (i.e., the residual) to determine whether the matching relationship is correct. However, the residual cannot effectively ensure the correctness of the match. In this paper, the Gaussian kernel method is introduced to map the one-dimensional indivisible residual to the high-dimensional feature space, and the inliers and the outliers are distinguished by fuzzy clustering. After simulation data and actual image data verification, the proposed algorithm has significant improvement in accuracy and efficiency compared with the traditional RANSAC algorithm. |
|---|---|
| ISSN: | 1682-1750 2194-9034 |