icp fitnessScore() and getFinalTransform() interpretation

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

icp fitnessScore() and getFinalTransform() interpretation

vijaykiran
Hi,

I'm new to using PCL. I'm trying to understand what icp fitness score and
getFinalTransform mean in terms of matching two point clouds.

Is the following interpretation correct?

- A higher fitness score means that the two point clouds I am trying to
match are distinctly different from each other? (For e.g., a hand and a ball
can give a fitness score of 600, but a ball and a sphere can give a fitness
score of 1e-08)

- If I have a source cloud and a target cloud, I should apply the
getFinalTransform matrix to the source cloud to align with the target.

Thanks,
Vijay.




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

Re: icp fitnessScore() and getFinalTransform() interpretation

MarkL
- Fitness Score:
Yes, the less it is, the better the match is.
http://docs.pointclouds.org/1.5.1/classpcl_1_1_registration.html#a647725f9c25170152a61e9ea26291f8a
According to documentation, it calculates the "sum of squared distances".
But in the code, you can see that it's actually the MSE
(http://docs.pointclouds.org/1.5.1/registration_8hpp_source.html).
You can also specify max_range, so that the score will only be calculated on
the inliers.

- Yes, it transforms the Source to the Target.


vijaykiran wrote

> Hi,
>
> I'm new to using PCL. I'm trying to understand what icp fitness score and
> getFinalTransform mean in terms of matching two point clouds.
>
> Is the following interpretation correct?
>
> - A higher fitness score means that the two point clouds I am trying to
> match are distinctly different from each other? (For e.g., a hand and a
> ball
> can give a fitness score of 600, but a ball and a sphere can give a
> fitness
> score of 1e-08)
>
> - If I have a source cloud and a target cloud, I should apply the
> getFinalTransform matrix to the source cloud to align with the target.
>
> Thanks,
> Vijay.
>
> _______________________________________________

> PCL-users@

>  / http://pointclouds.org
> http://pointclouds.org/mailman/listinfo/pcl-users





--
Sent from: http://www.pcl-users.org/
_______________________________________________
[hidden email] / http://pointclouds.org
http://pointclouds.org/mailman/listinfo/pcl-users