[robust_estimation] Add a GMS Filter (Grid-based Motion Statistics filter) [CVPR2017] #1049
Description
Add an implementation of [1]
- Implementation is based on the CVPR2017 paper code
Changes are:
-
rewrite the matcher in order to use some c++11 features (more compact implementation)
-
Fix error encountered when the points are out of the grid bounds
-
Use more explicit variable names
-
Add a Doxygen documentation
-
Add a unit test and a demo sample
[1] @inproceedings{bian2017gms,
title={GMS: Grid-based Motion Statistics for Fast, Ultra-robust Feature Correspondence},
author={JiaWang Bian and Wen-Yan Lin and Yasuyuki Matsushita and Sai-Kit Yeung and Tan Dat Nguyen and Ming-Ming Cheng},
booktitle={IEEE Conference on Computer Vision and Pattern Recognition},
year={2017}
}
Note: Since the motion are computed by tiny grid the more features the images have, the better the results will be (Left Distance ratio matches, Right Distance ratio matches + GMS filter).
SIFT_ULTRA:
openMVG_sample_image_matching_gmsfilter -t SIFT
1418 #Features on image A
3258 #Features on image B
366 #matches with Distance Ratio filter
228 #matches kept by the GMS Filter
AKAZE_ULTRA:
3422 #Features on image A
5454 #Features on image B
572 #matches with Distance Ratio filter
477 #matches kept by the GMS Filter
