How to use DenseFeatureDetector ?

asked 2015-03-04 04:37:33 -0500


I'm trying to obtain dense 3D reconstruction. Using SIFT or SURF detector/descriptor one can obtain only sparse reconstruction which can then be upgrated using dense matching (using DenseFeatureDetector). Unfortunately, I could not find any examples of its implementation. Could someone give me an example ? Should I rectify images before ?

Thank you

edit retag flag offensive close merge delete


I've used cv::Ptr< cv::FeatureDetector > detector = cv::FeatreDetector::create("Dense"); detector->detect(...), but I am not sure how you will use it? You may extract SIFT or SURF descriptors and then matching them with the ones from the 3D model?

thdrksdfthmn gravatar imagethdrksdfthmn ( 2015-03-04 06:49:49 -0500 )edit

Thanks for your answer, but I think that my question was formulated bad... Actually, the problem is not to apply DenseFeatureDetector, but to do matching using it. In the case of use of DenseFeatureDetector in 2 images, it creates a sort of grid on both, but it can't correctly find matches (I used SIFT descriptors). I think that some preliminary image treatment sould be done, like rectification or centering... So, I'm looking for existing examle where this technique has already been implemented.

Andrew26 gravatar imageAndrew26 ( 2015-03-04 06:59:46 -0500 )edit

OH, I think centering will be a nice thing, but I have no idea of how to do it.... Maybe some kind of detection?

thdrksdfthmn gravatar imagethdrksdfthmn ( 2015-03-04 08:18:12 -0500 )edit