difference between Depth Image and RGB Image

[Paul Doliotis]

Hi,
I am trying to capture RGB and Depth Video to run some experiments on
Sign Language.
I have noticed that the Depth Image is slightly different than the RGB
Image.
As if the Depth Image is more zoomed in or slightly scaled.

Ideally I would like to have absolute correspondences between pixels
in RGB and Depth images.
Any help on this?

Thanks!

Paul

[Zhaoyin Jia]

This page gives a detailed method of kinect calibration. After calibration I think they should align

http://nicolas.burrus.name/index.php/Research/KinectCalibration

--
Best

Zhaoyin Jia

[Marcos Slomp]

From the project's wiki: http://openkinect.org/wiki/Imaging_Information

"The RGB camera has a slightly larger angle of view than the Depth camera."

Since they (video and depth frames) have the same pixel resolution, a larger field-of-view causes the RGB camera to capture things with "less zoom".

Also note that the IR-receiver in the device has a small offset to the RGB camera.

The way to solve this is to find the correspondence is through some calibration.

Again, from the wiki: http://openkinect.org/wiki/Imaging_Information

"Lots of information on calibrating the depth camera is available on the ROS kinect_node page"

http://www.ros.org/wiki/kinect_node

I believe that Nicolas' RGB Demo also address the calibration issue.

http://nicolas.burrus.name/index.php/Research/KinectRgbDemoV4?from=Research.KinectRgbDemo

[Paul Doliotis]

thanks!
I have another question:
Is the calibration same for all kinects or you need different
parameters for each kinect?
I would assume that they are built in such a way that they are
identical. or no?

Paul

On Feb 7, 11:42 pm, Zhaoyin Jia <zhaoyin....@gmail.com> wrote:
> This page gives a detailed method of kinect calibration. After calibration I
> think they should align
>
> http://nicolas.burrus.name/index.php/Research/KinectCalibration
>

[Paul Doliotis]

thanks!
I have another question:
Is the calibration same for all kinects or you need different
parameters for each kinect?
I would assume that they are built in such a way that they are
identical. or no?

Paul

On Feb 7, 11:49 pm, Marcos Slomp <msl...@gmail.com> wrote:
> From the project's wiki:http://openkinect.org/wiki/Imaging_Information
> <http://openkinect.org/wiki/Imaging_Information>"The RGB camera has a
> slightly larger angle of view than the Depth camera."
> Since they (video and depth frames) have the same pixel resolution, a larger
> field-of-view causes the RGB camera to capture things with "less zoom".
> Also note that the IR-receiver in the device has a small offset to the RGB
> camera.
>

> <http://openkinect.org/wiki/Imaging_Information>The way to solve this is to

> find the correspondence is through some calibration.
> Again, from the wiki:http://openkinect.org/wiki/Imaging_Information
> "Lots of information on calibrating the depth camera is available on the ROS

> kinect_node page <http://www.ros.org/wiki/kinect_node>"http://www.ros.org/wiki/kinect_node
>
> I believe that Nicolas' RGB Demo also address the calibration issue.http://nicolas.burrus.name/index.php/Research/KinectRgbDemoV4?from=Re...

[Zhaoyin Jia]

I think they are not exactly identical, so you might need to calibrate each kinect. At least the online parameters do not work very well for my kinect.

--
Best

Zhaoyin Jia