[Ros-kinect] Rolling Shutter Compensation

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

[Ros-kinect] Rolling Shutter Compensation

Stefan Kohlbrecher
Hi everyone,

when moving Kinect while capturing images, rolling shutter effects are
visible in the RGB-images, I think one can also see them in the depth
images. This will become an issue once Kinect is used on (fast) moving
platforms (and can be very annoying when one has high frequency
vibration like e.g. on some quadrotors). Has anyone done tests on the
severity of these effects yet? Especially on platforms with low
computational power, compensating for rolling shutter effects after
publishing point clouds might be too expensive, so it would be most
efficient to have hooks for that in the driver when assembling point
clouds ..which of course isn't very nice decoupling wise. I thought I
might mention this possible issue early, will try Kinect on moving and
shaking platforms in the next weeks.

regards,
Stefan
_______________________________________________
Ros-kinect mailing list
[hidden email]
https://code.ros.org/mailman/listinfo/ros-kinect
Reply | Threaded
Open this post in threaded view
|

Re: [Ros-kinect] Rolling Shutter Compensation

Bill Morris
On Sat, 2010-11-20 at 11:36 +0100, Stefan Kohlbrecher wrote:

> Hi everyone,
>
> when moving Kinect while capturing images, rolling shutter effects are
> visible in the RGB-images, I think one can also see them in the depth
> images. This will become an issue once Kinect is used on (fast) moving
> platforms (and can be very annoying when one has high frequency
> vibration like e.g. on some quadrotors). Has anyone done tests on the
> severity of these effects yet? Especially on platforms with low
> computational power, compensating for rolling shutter effects after
> publishing point clouds might be too expensive, so it would be most
> efficient to have hooks for that in the driver when assembling point
> clouds ..which of course isn't very nice decoupling wise. I thought I
> might mention this possible issue early, will try Kinect on moving and
> shaking platforms in the next weeks.

I think we have most of the hardware and software to test this on a
quadrotor. We still need to work out a mount and a power adapter and
check how the current code runs on an Atom processor. We probably won't
get to this until after turkey day.

_______________________________________________
Ros-kinect mailing list
[hidden email]
https://code.ros.org/mailman/listinfo/ros-kinect