Calibration of depth measurement model for Kinect-type 3D vision sensors

Abstract

Accuracy of depth measurement with Microsoft Kinect and similar 3D vision sensors depends on variations in sensor production. Sensor reading may show significant systematic errors that can be compensated in software by using an adequate depth calibration model. This paper presents one such model and a procedure for identification of its parameters. An example calibration is given to illustrate the procedure and the attained improvements.