Calibration data structure for Kinect / PrimeSense sensors. More...

Public Attributes
CameraIntrinsicsParameters	depth_params

double	depth_to_rgb_quaternion [4]

double	depth_to_rgb_translation [3]

int	height

double	projector_depth_baseline

CameraIntrinsicsParameters	rgb_params

double	shift_offset

int	width

Detailed Description

Calibration data structure for Kinect / PrimeSense sensors.

This model is taken from the ROS wiki pages. TODO add a link here. Calibration parameters for this model can be fit by following the calibration procedure on the ROS wiki. TODO add a link here.

PrimeSense sensors (e.g., Kinect and PrimeSense) produce a disparity map that can be used to compute a dense depth map. For a given pixel in the IR camera, with pixel coordinates [u, v] and disparity d, the depth to the imaged object can be described as:

$z = \frac{8 * \mbox{projector\_depth\_baseline} * f}{(\mbox{shift\_offset} - d)}$

where $f$ is the IR camera focal length.

The $x$ and $y$ coordinates, in the IR camera frame, can be computed as:

$\begin{eqnarray*} x &=& (u - c_x) \frac{z}{f} \\ y &=& (v - c_y) \frac{z}{f} \end{eqnarray*}$

where $[c_x, c_y]$ is the center of projection for the IR camera.

Note that this coordinate convention is defined according to standard camera frames in computer graphics and vision, where +Z points along the optical axis away from the camera, +X is to the right, and +y is down. PrimeSense uses a different coordinate frame convention.