Keypoint matching and ORSA like algorithm
Hi all, I am working on image alignment using SIFT algorithm; but this also applies to SURF. Sometimes one needs to "refine" matching because of a few wrong match in the list. The ORSA algorithm looks adapted (Moisan-Stival ORSA (Optimized Random Sampling Algorithm) algorithm to filter out the false matches using the epipolar geometry constraint) but it uses only the X and Y coordinate of the keypoint and not the angle + orientation. Maybe using those informations it would be easier to spot those outliers, no ? Do you know an algorithm using those informations as well ? Thanks, -- Jérôme Kieffer <google@terre-adelie.org>
Hi Jerome On Wed, Sep 18, 2013 at 10:27 PM, Jérôme Kieffer <google@terre-adelie.org> wrote:
Sometimes one needs to "refine" matching because of a few wrong match in the list. The ORSA algorithm looks adapted (Moisan-Stival ORSA (Optimized Random Sampling Algorithm) algorithm to filter out the false matches using the epipolar geometry constraint) but it uses only the X and Y coordinate of the keypoint and not the angle + orientation. Maybe using those informations it would be easier to spot those outliers, no ?
Do you know an algorithm using those informations as well ?
Using RANSAC (included in skimage as well), you can define any criterium you choose for inliers. Regards Stéfan
participants (2)
-
Jérôme Kieffer -
Stéfan van der Walt