Dear Kilgus,
Thank you very much for your reply! I got a much better surface for online
data as your suggestion. KinectV2Controller (defined in
the ..\Modules\ToFHardware\KinectV2\mitkKinectV2Controller.h) provided
several good interfaces for accessing Kinect SDK v2 in MITK,
and ICoordinateMapper class (Kinect SDK interface,
http://msdn.microsoft.com/en-us/library/microsoft.kinect.kinect.icoordinatemapper.aspx
) was used to get 3D point from depth image and map coordinate from depth
to rgb image space. While I'm wondering if there is a way to set the camera
intrinsic parameters in Kinect SDK to improve the accuacry for 3D
reconstruction, like the issue with this link:
https://social.msdn.microsoft.com/Forums/en-US/e7346b37-204d-48c1-a204-14157719f653/internal-camera-parameters-and-kinect-fusion?forum=kinectsdk
Regards,
Deqiang
2014-10-20 18:09 GMT+08:00 Kilgus, Thomas <[email protected]>:
> Hi Deqiang Xiao,
>
>
>
> please make sure you keep the conversation on the list. For some reason
> only I got the mail. Please consider alternatives for huge data files
> instead of attaching them. Maybe your mail got rejected due to the size?
>
>
>
> The Kinect v2 implementation is somewhat experimental. I was not really
> aware that users are already testing this feature. But your feedback is
> very welcome!
>
>
>
> Regarding your questions:
>
> 1. How did you generate the MITK surface? The Microsoft SDK does
> some magic, which I didn’t understand yet. If you generate the surface via
> our player, the reconstruction of the Kinect v1 will be used, which is
> wrong for Kinect v2 data. If you plugin an actual Kinect v2, the SDK code
> should be used and surfaces should be exactly the same as yours.
>
> 2. See 1.
>
> 3. You can have a look at the controller code where I access the
> Kinect SDK [1]. Feel free to play with it. I pass the SDK data to the
> workbench via properties, which is certainly not perfect. Any contributions
> are welcome. It would be great to find a smoother solution.
>
> 4. It is an experimental feature tested only by me. The surfaces
> used to look much better than the one you send me.
>
>
>
> Unfortunately, I don’t have a PC with Windows 8 and USB 3.0 right now to
> test the code again.
>
>
>
> Regards,
>
> Thomas
>
>
>
>
>
> [1]
> http://mitk.org/git/?p=MITK.git;a=blob;f=Modules/ToFHardware/KinectV2/mitkKinectV2Controller.cpp;h=a504bd068bcc726a4c8ff395c6deec8d4f202849;hb=HEAD
>
>
>
>
>
>
>
>
>
> *Von:* Deqiang Xiao [mailto:[email protected]]
> *Gesendet:* Samstag, 18. Oktober 2014 13:03
> *An:* Kilgus, Thomas
> *Cc:* [email protected]
> *Betreff:* Problems about MITK ToF Plugin working with Kinect v2 (Kinect
> for windows v2)
>
>
>
> Dear Mr. Kilgus,
>
>
>
> I got the latest code of MITK and complied ToF plugin through Visual
> Studio 2008 to connect Kinect for windows v2 successfully.
>
> While there are some problems about the RGBD data achieved from ToF plugin:
>
>
>
> 1) The surface generated from ToF plugin looks like wrong when compared
> with that generated through Kinect SDK v2 (Kinect SDK provided many
> interfaces for processing RGBD data from Kinect sensor on Windows platform),
>
> Two surface files (*.stl) generated from MITK-ToF and Kinect SDK based on
> the same depth image can be found in the attachment. By the way, the camera
> intrinsic parameters are already updated before data acquiring.
>
>
>
> 2) The RGB and depth image are mirror flipped, and MITK will crash down
> once it reconstructs surface from depth image during RGB-D image acquiring.
>
>
>
> 3) Can I call the functions from Kinect SDK in MITK-ToF? Cause I wanna to
> get the corresponding relationship of pixels between RGB and depth image
> from Kinect v2 (The resolutions of these two images are different, then I
> cannot map the pixel from RGB to depth directly).
>
>
>
> 4) Did your team test MITK-ToF with Kinect for Xbox one or Kinect for
> windows v2? How about it works?
>
>
>
> Any help is appreciated!
>
>
>
> Regards,
>
> Deqiang Xiao (PhD student)
>
>
>
> Research Lab for Medical Imaging and Digital Surgery
>
> Shenzhen Institutes of Advanced Technology
>
> Chinese Academy of Sciences
>
> Add. : No. 1068, Xueyuan Boulevard, Xili University Town
>
> Nanshan, Shenzhen, 518055
>
> P.R. China
>
------------------------------------------------------------------------------
Comprehensive Server Monitoring with Site24x7.
Monitor 10 servers for $9/Month.
Get alerted through email, SMS, voice calls or mobile push notifications.
Take corrective actions from your mobile device.
http://p.sf.net/sfu/Zoho
_______________________________________________
mitk-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/mitk-users
