regarding curvature computation in 3dSIFTkeypoint detector

classic Classic list List threaded Threaded
5 messages Options
Reply | Threaded
Open this post in threaded view
|

regarding curvature computation in 3dSIFTkeypoint detector

shome
In "COMPARISON OF 3D INTEREST POINT DETECTORS AND DESCRIPTORS
FOR POINT CLOUD FUSION: R. H ̈ansch a, ∗ , T. Weber a , O. Hellwich a "

it is mentioned in sec 3.2 that:

The Scale Invariant Feature Transform (SIFT, (Lowe, 2004)) originally developed for 2D images was adapted by the community of the Point Cloud Library (PCL) to 3D point clouds (Rusu and Cousins, 2011) by replacing the role of the intensity of an pixel in the original algorithm by the principal curvature of a point within the 3D cloud.

So do we need to estimate curvature at every point in the pointcloud before we do sift computation.
Reply | Threaded
Open this post in threaded view
|

Re: regarding curvature computation in 3dSIFTkeypoint detector

andersgb1
As far as I can see, yes: http://docs.pointclouds.org/trunk/sift__keypoint_8h_source.html#l00044. You get these curvatures (which are not the principal curvatures btw.) by the NormalEstimation for free. Additionally, in the absence of curvature, it seems to use intensity/color instead.

On Fri, Aug 28, 2015 at 8:05 AM, shome <[hidden email]> wrote:
In "COMPARISON OF 3D INTEREST POINT DETECTORS AND DESCRIPTORS
FOR POINT CLOUD FUSION: R. H ̈ansch a, ∗ , T. Weber a , O. Hellwich a "

it is mentioned in sec 3.2 that:

The Scale Invariant Feature Transform (SIFT, (Lowe, 2004)) originally
developed for 2D images was adapted by the community of the Point Cloud
Library (PCL) to 3D point clouds (Rusu and Cousins, 2011) by replacing the
role of the intensity of an pixel in the original algorithm by the
*principal curvature* of a point within the 3D cloud.

So do we need to estimate curvature at every point in the pointcloud before
we do sift computation.



--
View this message in context: http://www.pcl-users.org/regarding-curvature-computation-in-3dSIFTkeypoint-detector-tp4039251.html
Sent from the Point Cloud Library (PCL) Users mailing list mailing list archive at Nabble.com.
_______________________________________________
[hidden email] / http://pointclouds.org
http://pointclouds.org/mailman/listinfo/pcl-users


_______________________________________________
[hidden email] / http://pointclouds.org
http://pointclouds.org/mailman/listinfo/pcl-users
Reply | Threaded
Open this post in threaded view
|

Re: regarding curvature computation in 3dSIFTkeypoint detector

shome
i am looking for geometric representation.so the rgb/intensity thing may not work for me.
So what you are suggesting is that i create the 3d_sift_detector as below:

pcl::SIFTKeypoint<pcl::PointNormal, pcl::PointXYZ> sift_detector;

and the let the sift detector detect the curvature field itself using
SIFTKeypointFieldSelector<PointInT> getFieldValue_;

Reply | Threaded
Open this post in threaded view
|

Re: regarding curvature computation in 3dSIFTkeypoint detector

andersgb1
No, it doesn't detect the curvature itself, it only uses it. So you have to run your XYZ data through NormalEstimation first such that you get out of that a filled PointNormal cloud, and then you can run that through the SIFT detector.

At least, so it seems to me - I've never actually used this class before :)

On Fri, Aug 28, 2015 at 9:11 AM, shome <[hidden email]> wrote:
i am looking for geometric representation.so the rgb/intensity thing may not
work for me.
So what you are suggesting is that i create the 3d_sift_detector as below:

pcl::SIFTKeypoint<pcl::PointNormal, pcl::PointXYZ> sift_detector;

and the let the sift detector detect the curvature field itself using
SIFTKeypointFieldSelector<PointInT> getFieldValue_;





--
View this message in context: http://www.pcl-users.org/regarding-curvature-computation-in-3dSIFTkeypoint-detector-tp4039251p4039254.html
Sent from the Point Cloud Library (PCL) Users mailing list mailing list archive at Nabble.com.
_______________________________________________
[hidden email] / http://pointclouds.org
http://pointclouds.org/mailman/listinfo/pcl-users


_______________________________________________
[hidden email] / http://pointclouds.org
http://pointclouds.org/mailman/listinfo/pcl-users
Reply | Threaded
Open this post in threaded view
|

Re: regarding curvature computation in 3dSIFTkeypoint detector

shome
In reply to this post by shome
yah dats what i meant. using precomputed normals.