An extrinsic calibration method for 3D range surface sensors: an application in radiotherapy patient positioning

The use of 3D surface sensor has been recently proposed for patient positioning in radiotherapy treatments. In this kind of applications, relating the coordinate system of 3d sensor with the coordinate system of treatment's room is crucial and needs a sensor's calibration. Sensor's calibration depends of its measuring principle; so calibrating a time-of-fly sensor has some differences to calibrating an active triangulation sensor. Additionally the intrinsic calibration requires knowing the internal parameters of sensor, which are frequently unknown for final user. This work presents a method for calibrating 3D surface sensor independently of its measuring principle. It uses directly the cloud of points to retrieve the extrinsic parameters, relating the virtual sensor space with a real coordinate system (three laser's beams of a radiotherapy room). A comparison between two different calibrations objects is presented. The use of 3D points allows a flexible, accurate and robust extrinsic calibration. Tests results show that it is possible to relate two coordinates systems with less of 1.2 mm error