Literature/Explanation for Harris 3D Keypoints

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

Literature/Explanation for Harris 3D Keypoints

hanno
Hi,

I've seen the implementation of the 3D versions of
Harris/Noble/Lowe/Tomasi and Curvature-based keypoint detection
methods. Is there some literature describing these methods (exactly
these versions, for 3d pointclouds)? Or is it new, unpublished work?

What I get from reading the code, is that the x-value of the normals
in the neighborhood of a point is used to calculate a covariance
matrix, which is then used to compute the keypoint response
(intensity). Is that correct?

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

Re: Literature/Explanation for Harris 3D Keypoints

Suat Gedikli
Hi,

as far as we know there is no publication about this. The idea is basically taken from the 2D case, where image the covariance matrix over the gradients is computed and the response is obtained from that.
For the 3D case, we just replaced the gradients by the surface normals and use the same responses.

-Suat

On Mon, Dec 19, 2011 at 3:21 PM, Hanno Jaspers <[hidden email]> wrote:
Hi,

I've seen the implementation of the 3D versions of
Harris/Noble/Lowe/Tomasi and Curvature-based keypoint detection
methods. Is there some literature describing these methods (exactly
these versions, for 3d pointclouds)? Or is it new, unpublished work?

What I get from reading the code, is that the x-value of the normals
in the neighborhood of a point is used to calculate a covariance
matrix, which is then used to compute the keypoint response
(intensity). Is that correct?

Greetings,
Hanno
_______________________________________________
[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: Literature/Explanation for Harris 3D Keypoints

hanno
Ok, thank you for your answer!
Reply | Threaded
Open this post in threaded view
|

Re: Literature/Explanation for Harris 3D Keypoints

raphael favier
Hi there,

is there a reason why the pcl::HarrisKeypoint3D class does not appear in the modules Keypoints of the documentation (1.4 and 1.5) ?

http://docs.pointclouds.org/1.4.0/group__keypoints.html

Raph
Reply | Threaded
Open this post in threaded view
|

Re: Literature/Explanation for Harris 3D Keypoints

Radu B. Rusu
Administrator
Fixed in trunk. Thanks! :)

Cheers,
Radu.
--
http://pointclouds.org

On 01/17/2012 07:45 AM, raphael favier wrote:

> Hi there,
>
> is there a reason why the pcl::HarrisKeypoint3D class does not appear in the
> modules Keypoints of the documentation (1.4 and 1.5) ?
>
> http://docs.pointclouds.org/1.4.0/group__keypoints.html
>
> Raph
>
> --
> View this message in context: http://www.pcl-users.org/Literature-Explanation-for-Harris-3D-Keypoints-tp3599877p3666494.html
> Sent from the Point Cloud Library (PCL) Users 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: Literature/Explanation for Harris 3D Keypoints

luis_alex
In reply to this post by Suat Gedikli
Hi

What about this publication?
https://dl.acm.org/citation.cfm?id=2134343
http://www.dcc.uchile.cl/~isipiran/papers/SB11b.pdf

Cheers,
Luis


Suat Gedikli wrote
Hi,

as far as we know there is no publication about this. The idea is basically
taken from the 2D case, where image the covariance matrix over the
gradients is computed and the response is obtained from that.
For the 3D case, we just replaced the gradients by the surface normals and
use the same responses.

-Suat

On Mon, Dec 19, 2011 at 3:21 PM, Hanno Jaspers <[hidden email]
> wrote:

> Hi,
>
> I've seen the implementation of the 3D versions of
> Harris/Noble/Lowe/Tomasi and Curvature-based keypoint detection
> methods. Is there some literature describing these methods (exactly
> these versions, for 3d pointclouds)? Or is it new, unpublished work?
>
> What I get from reading the code, is that the x-value of the normals
> in the neighborhood of a point is used to calculate a covariance
> matrix, which is then used to compute the keypoint response
> (intensity). Is that correct?
>
> Greetings,
> Hanno
> _______________________________________________
> [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: Literature/Explanation for Harris 3D Keypoints

fedassa
Hi,

what instead about the implementation of Harris6d detector? how does it use together intensity and shape information? does it merge somehow together the covariance matrix computed on intensity gradients with that computed on normals?

Fede
Reply | Threaded
Open this post in threaded view
|

Re: Literature/Explanation for Harris 3D Keypoints

Sombra
In reply to this post by luis_alex
I'm working with this and I suppose that Suat Gedikli hasn't used Sipiran and Busto's version. I'm actually trying to implement that one, and I will try to post it to developers if it can be accepted. At this moment I don't know the requirements to contribute to PCL.
Reply | Threaded
Open this post in threaded view
|

Re: Literature/Explanation for Harris 3D Keypoints

Jochen Sprickerhof
Administrator
Hi Sombra,

* Sombra <[hidden email]> [2013-10-31 05:59]:
> I'm working with this and I suppose that Suat Gedikli hasn't used Sipiran and
> Busto's version. I'm actually trying to implement that one, and I will try
> to post it to developers if it can be accepted. At this moment I don't know
> the requirements to contribute to PCL.

Sounds like a great addition. There are no requirements, apart from that
it needs to be BSD licenced and fit the PCL style. Just send the code as
a pull request on Github and we can figure out the details.

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

Re: Literature/Explanation for Harris 3D Keypoints

nizar sallem
In reply to this post by Sombra
Could you be more specific about Sipiran and Bustos ?

Cheers,
--
Nizar

On 31/10/2013 13:59, Sombra wrote:
> Sipiran and
> Busto's
_______________________________________________
[hidden email] / http://pointclouds.org
http://pointclouds.org/mailman/listinfo/pcl-users
Reply | Threaded
Open this post in threaded view
|

Re: Literature/Explanation for Harris 3D Keypoints

nizar sallem
In reply to this post by Sombra
Well if you are talking abou this article "Harris 3D: interest point
detection on meshes" then it is really mesh oriented and the novelty in
Suat's work is that it doesn't require point cloud to be organised or
the the mesh information to be available.
Indeed, in sipiran work he relies on rings information to organise the
cloud and mesh resolution to find neighbors so that would be another
keypoint detector probably better implement it using PolygonMesh ;)

Cheers,
--
Nizar

On 31/10/2013 13:59, Sombra wrote:

> I'm working with this and I suppose that Suat Gedikli hasn't used Sipiran and
> Busto's version. I'm actually trying to implement that one, and I will try
> to post it to developers if it can be accepted. At this moment I don't know
> the requirements to contribute to PCL.
>
>
>
> --
> View this message in context: http://www.pcl-users.org/Literature-Explanation-for-Harris-3D-Keypoints-tp3599877p4030507.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