Semi-automatic calibration of 3D camera images — Monitoring activities made easy

A major practical problem when implementing home monitoring using 3D camera technology is the calibration process which serves to relate the camera coordinate system to a coordinate system interpretable by the caregiver. In this paper, we propose a calibration algorithm which can relate both coordinate systems without performing actual measurements of the position of reference points in the room. The calibration tool is constructed such that it can be operated by any caregiver and hence eases the deployment of the home monitoring appliances. Our method (semi)automatically detects a plane corresponding to the ground floor in the three dimensional images. Random points from that plane are used to calculate rotation angles in a least squares manner. Experiments are performed to investigate how the obtained results depend on the parameters of the algorithm.