On the '2016 R2' and '2016 R3' SDKs, 2D depth image coordinates are converted to 3D world coordinates with the instruction ProjectDepthToCamera.
If you wish to track the face specifically in the R2 or R3 SDK, there are 78 face 'points of interest' in the SDK's facial system whose position is provided in image and world coordinates.
In SDK 2.0, the instruction rs2_deproject_pixel_to_point is used to convert 2D image points to 3D world points.
Maybe my expression was not clear.
I want to know how to get the Oc-XcYcZc of SR300(c means camera)：Origin coordinates and x, y, z, axis vector.
I will do a registration between the camera three-dimensional coordinate system with the three-dimensional human face coordinate system build by my self,so i want to know the Oc-XcYcZc of SR300.