real world coordinates and field of view of kinect v2

Question

Hi,

I have read some different specs on kinect v2 regarding fov I have 60 in vertical and 70.6 in horisontal. Can anyone point me to the right source to get this data? Or confirm this.

I am retrieving a depth data file and then I process it in matlab and doing the mapping from depth data to x-y-coordinates like in this post:

http://stackoverflow.com/a/18199938

Can any one confirm deny if this is _the_ correct way to do the mapping? Obviously with the gov specs of 60*70.6. It seems like it is correct when I do some testing of objects but I need to know for sure.

Greatful for comments, thanks!

Answer 1

Every source has a FrameDescription object. This class/object contains a Vertical/Horizontal field of view property. We recommend you never hard code these into your code and pull the data from this object.

https://msdn.microsoft.com/en-us/library/windowspreview.kinect.framedescription.aspx?f=255&MSPPError=-2147217396

---

Carmine Sirignano - MSFT

Answer 2

Thanks, I will see if I can manage to use the FrameDescription object. But HorizontalFieldOfView and VerticalFieldOfView should always be the same, right? I am doing all the processing in matlab from just taking out a textfile of depth data by manipulating the c++ code given in kinect's software examples. D2D-depth basics I think that one is called!

Is there any way I can get point's real world xyz- coordinates directly? The depth information I currently obtain is 512*424 _pixels_ with z-coordinates (depth)?

Answer 3

I would not make that assumption as the API's were designed to enable that type of change. To get real-world coordinates, you need to map that depthframe to cameraspace.

https://msdn.microsoft.com/en-us/library/microsoft.kinect.coordinatemapper.mapdepthframetocameraspace.aspx

---

Carmine Sirignano - MSFT