Using Kinect for Xbox One with MITK-ToF
Note: The Kinect for Xbox One was experimentally implemented in MITK-ToF, but still has some issues. Contributions to improve the support are always welcome.
What does work? If you follow all the steps you should be able to acquire depth and color data with MITK-ToF and also generate surfaces.
What does not work? The reconstruction of the depth data seems to be different from other depth sensors such as ToF-cameras or the Kinect for Xbox 360. If you use the classic MITK-ToF reconstruction, the surface will be flawed. The Microsoft SDK provides a function to efficiently generate correct surfaces from the depth image. This data can be streamed into the datastorage of MITK, however, this means that the preprocessing pipeline of MITK is skipped and that reconstructing the correct surface from recorded data is not possible.
First step: The Kinect for Xbox One needs Windows 8 with a fully supported USB 3.0 port. Make sure to run a Microsoft SDK example and check that your Kinect is principally working on your machine before doing any effort with MITK. Experience showed that this is often not the case. Here you can find an installation guide: https://www.microsoft.com/en-us/kinectforwindows/purchase/sensor_setup.aspx
Installation for MITK-ToF: The installation for MITK-ToF is rather simple and straight forward:
- Download and install the Microsoft SDK: https://www.microsoft.com/en-us/download/details.aspx?id=44561
- Activate the CMake flag: MITK_USE_TOF_KINECTV2 and set the MITK_KINECTV2_INCLUDE_DIR and MITK_KINECTV2_LIB
- Now you should be able to connect the camera with, for instance, the tofutil plugin. If you generate the surface while the actual camera is connected, the data should be "correct" (meaning Microsoft's reconstruction is used). Preprocessing does not work, as mentioned before.