[Ros-kinect] Kinect depth discretization

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

[Ros-kinect] Kinect depth discretization

Ivan Dryanovski
Hi everyone,

According to the Openni camera documentation, the z value in the depth
image corresponds to the distance in meters. However, when inspecting
the data, you can see that the z values followin discrete jumps.

Back when I was working with the libfreenect driver (and its ros
wrapper), the output of the Kinect was a 11-bit number, which was
linear with the inverse of the distance in meters. This produced
discretization levels with steps that increased as the distance to the
object increased. Is the same true for the openni driver? Is it still
using 11 bits internally, and are they still related to the meter
distance by the same function? I tried looking through the wrapper
source, but it's not there.

The reason I'm asking this is because I'm writing an application where
it might be advantageous if I could work with the data in its raw
(unlinearized) discrete values, rather than the meter distance.

Thank you,
Ivan Dryanovski
_______________________________________________
Ros-kinect mailing list
[hidden email]
https://code.ros.org/mailman/listinfo/ros-kinect
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: [Ros-kinect] Kinect depth discretization

Suat Gedikli
Hi,

the steps come from the fact, that the Primesense/Kinect uses just 1/8 of a pixel quantization for the disparity image, which is used to determine the depth in the OpenNI driver by primesense.

Suat
Suat
Loading...